Skip to main content

Full text of "soil analysis"

See other formats

The Impact of Agroforestry-Based Soil Fertility 
Replenishment Practices on the Poor in Western Kenya 

Frank Place 
Michelle Adato 
Paul Hebinck 
and Mary Omosa 




sustainable solutions for ending hunger and poverty 

jcppi® World Agroforestry Centre 


Copyright © 2005 International Food Policy Research Institute. All rights reserved. 
Sections of this material may be reproduced for personal and not-for-profit use 
without the express written permission of but with acknowledgment to IFPRI. To 
reproduce the material contained herein for profit or commercial use requires express 
written permission. To obtain permission, contact the Communications Division 

International Food Policy Research Institute 

2033 K Street, NW 

Washington, DC 20006-1002 USA 

Telephone +1-202-862-5600 

Library of Congress Cataloging-in-Publication Data 

The impact of agroforestry-based soil fertility replenishment practices 
on the poor in western Kenya / Frank Place . . . [et al.]. 

p. cm. — (Research report ; 142) 
Includes bibliographical references. 

ISBN 0-89629-144-8 (alk. paper) 

1. Soil fertility — Kenya. 2. Agroforestry extension — Kenya. 3. Agroforestry 
projects — Kenya. I. Place, Frank, Dr. II. Research report (International Food 
Policy Research Institute) ; 142 
S591.55.K4I47 2005 
631.4'22'096762— dc22 2005013830 


List of Tables iv 

List of Figures vi 

List of Boxes vii 

Foreword viii 

Acknowledgments x 

Summary xi 

1. Introduction 1 

2. Research Methods 6 

3. The Context of the Research 21 

4. Poverty in Context 27 

5. Household-Level Livelihood Strategies and Their Context 34 

6. Processes and Patterns of Adoption 41 

7. Soil Fertility Replenishment and Rural Peoples' Livelihood 

in Western Kenya 61 

8. Dissemination of Soil Fertility Replenishment Technologies: 

Comparing Approaches, Methods, and Experiences 81 

9. Human and Social Capital Formation: Dissemination within the Villages 103 
10. Conclusions and Recommendations 125 
Appendix A: Results and Tests from First-Stage Regressions 132 
Appendix B: Six Village-Level Case Studies of Dissemination Processes 136 
References 165 



1.1 Phase 1, Wave 1 case studies of impact of agricultural research under the 

IFPRI/SPIA project 2 

2.1 Case study villages 14 

2.2 Villages and dissemination approaches in the dissemination study 16 

2.3 Research design matrix: Assets, vulnerability, and livelihoods 17 

2.4 Research design matrix: Dissemination strategies 19 

4.1 Distribution of poverty in pilot villages (n = 104) using 

alternative classifications 32 

4.2 Distribution of poverty in non-pilot villages (n = 360) using 

alternative classifications 32 

5.1 Livelihood strategies pursued by individuals in pilot villages (130 households) 35 

6.1 Use of agroforestry in the pilot villages over time 

(as percentage of 1,538 households) 42 

6.2 Use of agroforestry in non-pilot villages over time 

(as percentage of 360 households) 43 

6.3 Patterns of use of improved fallows and biomass transfer in the pilot villages 

(as percentage of 1,598 households) 43 

6.4 Rates of use of improved fallows in early non-pilot area villages 44 

6.5 Size of fallows over time in pilot villages (square meters) 45 

6.6 Planting of tithonia biomass transfer systems on farm over time in pilot 

villages (as percentage of households planting) 45 

6.7 Household factors related to adoption of improved fallows in pilot villages, 
1997-2001 (n = 1,583) 48 

6.8 Household factors related to adoption of biomass transfer in pilot villages, 
1997-2001 (n = 1,583) 49 

6.9 Multinomial logit results for adoption of improved fallows in non-pilot 

villages (n = 361) 51 

6.10 Multinomial logit results for adoption of biomass transfer in non -pilot 

villages (n = 361) 52 

6.11 Use rates of soil fertility management options over time in non-pilot 

project areas 59 



7.1 Soil fertility practices and maize yield impacts 66 

7.2 Description of household liquid assets in pilot and non-pilot villages 71 

7.3 Econometric results from second-stage regression of agroforestry on changes 

in assets in pilot villages (n = 97) 72 

7.4 Total non-food expenditures, per capita non-food expenditures, and changes 

during the three-month-long rainy season in 2000 and 2002 (in U.S. dollars) 73 

7.5 Econometric results from second-stage regressions of agroforestry on 
changes in non-food expenditures and per capita non-food expenditures 

in pilot villages (n = 102) 74 

7.6 Percentage of daily requirements of nutritional measures at the household 

level prior to a long -rain harvest 75 

7.7 Econometric results from second-stage regression of agroforestry use on 
nutritional measurements (n = 102) 76 

8.1 Village selection for dissemination study 82 

8.2 Summary of reach and effectiveness of different sources of information on 

SFR (as percentage of all households located in relevant villages) 93 

8.3 Percentage of households receiving information on agroforestry from other 

farmers or any source 94 

8.4 Percentage of households with direct contact for SFR information, by source 

and wealth group 96 

9.1 Knowledge gain for tithonia and improved fallows in the six study villages 

from focus group discussions 119 

9.2 Knowledge acquisition in agricultural topics from household survey in 

non-pilot villages (« = 361) 120 

A.l OLS regression results for improved fallow area 132 

A.2 Tobit regression results for improved fallow area 133 

A.3 OLS regression results for number of seasons practicing biomass transfer 133 

A.4 Tobit regression results for number of seasons practicing biomass transfer 134 


2.1 Sustainable livelihoods framework 8 

6.1 Adoption patterns of improved fallows and biomass transfer in the pilot 

villages over time, 1997-2001 (as percentage of 1,630 households) 43 

6.2 Adoption patterns of improved fallows and biomass transfer in the pilot 

villages by 2001 (as percentage of 1,630 households) 44 

8.1 Village map of institutions involved with SFR information exchange — 
Bukhalalire village, poor men 83 

8.2 Village map of institutions involved with SFR information exchange — 

Mutsulio village, poor women 84 

8.3 Village map of institutions involved with SFR information exchange — 

Mutsulio village, poor men 85 

8.4 Average preferences for dissemination methods used by external 

organizations in six villages 100 

9.1 Average assessments of relative importance of internal disseminators in poor 

versus non-poor focus groups 105 

9.2 Average assessments of relative importance of internal disseminators in men's 

versus women's focus groups 106 

9.3 Sauri poor women's group: At least four years of primary education 115 

9.4 Sauri non-poor women's group education level: At least four years of 

primary education 116 

9.5 Mutsulio village — poor women: At least four years of primary education 117 

9.6 Mutsulio village — poor men: At least four years of primary education 118 

9.7 Average knowledge gain for each technology, by village 119 



5.1 Relay Type of Strategies 36 

5.2 Types of Farmers and Farming Systems 38 

7.1 Increased Farm Yields 62 

7.2 Mitigating Vulnerability 64 



For decades, there has been significant investment in the development of agricultural 
technologies that aim to increase productivity of smallholder farms in Africa. At a 
macro-level, however, farm output and productivity have stagnated and poverty rates 
have remained stubbornly high, even increasing in some areas. 

It is widely acknowledged that policy and infrastructural constraints play a large role in 
reducing incentives for farmers to invest in agriculture. Yet the fact that farmers have made 
some investments and that some progress has occurred suggests that characteristics of the 
technologies themselves, or the way in which they are promoted, also facilitate or inhibit wider 
adoption and impact. 

This research report, part of a set of studies on the impact of agricultural research on 
poverty led by IFPRI, analyzes the adoption and impact of agroforestry techniques for soil fer- 
tility enhancement in one of the poorest regions of the world — the western Kenyan highlands. 
It further examines the role that government and nongovernmental organizations and their dif- 
ferent dissemination methods play in reaching potential users and helping them understand 
and use this knowledge-intensive technology. In this study, the researchers have used quanti- 
tative and qualitative research methods to make discoveries and develop insights that neither 
method alone could accomplish. 

The authors find that improved fallows and biomass transfer systems are attractive to the 
poor because they are low in cost and provide noticeable increases in crop yields. Lower and 
higher income groups in the study villages use these systems in similar ways — this is not the 
case for fertilizer use — but the small farm sizes of the region limit the impact of this technol- 
ogy. The size of area under these systems remains small after six years of dissemination, in- 
dicating that yield improvements do not translate into significant household-level welfare 
impacts for the most part. 

Maintaining information flows is a challenging task with such high rates of poverty and 
the continuous search for livelihoods on and off farm. Persons in close contact with develop- 
ment organizations increase their knowledge of soil fertility management, but villagers noted 
problems with the quantity and quality of information. Different methods for disseminating 
knowledge have strengths and weaknesses with respect to reaching poor farmers and women, 
and this has implications for social capital. 

It is challenging to reduce short-term poverty rates in highly populated areas where farm 
sizes have decreased to less than one hectare. Adverse shocks are ubiquitous and they almost 
always deplete household asset bases. The authors do not find any single occupation or in- 
vestment that always improves welfare. Thus, the study confirms in a clear way that poverty 
reduction will require sets of interventions and greater understanding of their sequencing and 
integration. Low-cost methods for raising soil fertility can be coupled with other feasible en- 
terprises for poor farmers, such as the use of improved maize varieties that are resistant to 
streak virus and the increased planting of higher value crops like kales and climbing beans. 
Greater investment in poultry or ruminants and in fruits or woodlots are also within the capacity 



of poor households. The spread and speed of such investments would need to be underpinned 
by increased access to capital or credit, which is very scarce in the study site. 

Joachim von Braun 
Director General, IFPRI 


The authors acknowledge the central contributions of members of the research team: 
Wesley Ongadi and Pamella Opiyo lived in several of the research villages for six 
months, carrying out the case study research; Mary Nyasimi helped supervise the dis- 
semination focus groups and designed figures of the village maps and knowledge ladders; 
Maggie Lwayo, Edward Ontita, and Christopher M. C. O'Leary assisted with data analysis. 

The authors also thank Anthony Bebbington, Jere Behrman, and Robert Chambers for their 
many valuable contributions as members of the project's Independent Advisory Committee. 
Finally, they thank Lawrence Haddad, Ruth Meinzen-Dick, and John Pender for helpful com- 
ments on an earlier draft of this report; Ginette Mignot for project administration; and Carole 
Douglis and Jay Willis for editorial assistance. 

Funding for the project was provided by the United Kingdom's Department for Inter- 
national Development (DID); the Standing Panel on Impact Assessment (SPIA) of the Con- 
sultative Group on International Agricultural Research (CGIAR); the Australian Centre 
for International Agricultural Research (ACIAR); the Danish International Development Agency 
(Danida); the Government of the Netherlands; the International Fund for Agricultural Devel- 
opment (IFAD); and the Food and Agriculture Organization of the United Nations (FAO). 


Western Kenya is one of the most densely populated areas in Africa. Farming there 
is characterized by low inputs and low crop productivity. Poverty is rampant in the 
region. Yet the potential for agriculture is considered good. In the study described 
here, researchers looked specifically at soil fertility replenishment (SFR) systems as part of a 
larger IFPRI effort to examine the impact of agricultural research on reducing poverty. Fo- 
cused on two specific systems — the tree- based "improved fallow" system and the biomass 
transfer system — the study compared rates of adoption in poor and nonpoor communities and 
evaluated the extent to which their adoption reduced poverty. 

Improved fallow refers to the intentional planting of a fallow species. Improved fallows 
are more efficient than natural fallows, typically achieving the same effect on crop produc- 
tivity in a much shorter time. Biomass transfer systems are those in which organic nutrient 
sources are grown in one place and then transferred to crops in another place. Such a system 
allows farmers to grow crops continuously, an advantage over the improved fallow system. 
The space available for producing organic nutrient sources on-farm is limited, however. 

Combining Qualitative and Quantitative Analysis to 
Assess Poverty 

The range of issues covered in the study required the use of a variety of research methods and 
interdisciplinary perspectives. For example, distinguishing the poor from the nonpoor for the 
purposes of the study was not straightforward. The researchers used a variety of methods to 
assess poverty levels, including quantitative measures from surveys, enumerator ratings, and 
farmer self-assessments. Local perceptions of poverty and the role of gender, power, and other 
social constructs in understanding why and how farmers adopted the new systems, and the 
impact of the systems, could only be explored using qualitative research methods and socio- 
logical perspectives. These were combined with quantitative measures of adoption and impact, 
along with economic analysis. The researchers also drew on long-term knowledge of the re- 
gion based on work by members of the study team and others. 

Study Findings 

• Over the course of the study, welfare or livelihood outcomes worsened for many house- 
holds. There was a general deterioration in welfare indicators, including assets, expendi- 
tures, and food consumption. Particularly striking was that households with relatively high 
welfare indicators in the initial period suffered the greatest losses. This was due partly to 
the large number of adverse shocks affecting households and the cultural obligations felt 
by all community members. 

• Households did see the importance of SFR, and there were many human capital impacts. 
Both the qualitative and quantitative research found significant knowledge acquisition 
taking place, not only for agroforestry methods but also for general soil management and 
farming practices. People valued this information and often put it into practice. 



• The poor adopted SFR strategies at the same rate as the nonpoor. Adoption rates were 
not outstanding but were encouraging, with about 20 percent of all farmers using the 
technologies on a regular basis. 

• Adoption at the early stage was at low levels of intensity. Although an encouraging 
number of households used or tested SFR practices, the size of plots on which they were 
applied was small. It is not yet known whether this is a ceiling or a consequence of the 
early stage of dissemination. 

• The dissemination analysis found that farmers appreciated some aspects of different dis- 
seminating organizations and the many different methods tried. Although characteristics 
of SFR affected whether people adopted a system, aspects of the dissemination process 
also affected adoption. The dissemination analysis found that the main feature of most 
dissemination approaches — group-based methods — can strengthen human and social 
capital, and that farmers of different social status benefited from such methods. How- 
ever, this analysis also found that group-based approaches may disadvantage farmers of 
lower social status and women, who are less likely to participate in or dominate groups. 
These findings reinforce the idea that it is best to use a variety of methods to disseminate 
new technologies or knowledge. 

• Sustainability of dissemination structures and processes proved possible, but challenging, 
due to problems encountered by farmer groups, limited capacity of local administration, 
social dynamics within villages, and limited cost-sharing ability. Monitoring would help 
to pick up these problems so that resolutions could be sought where possible. 

• SFR did significantly raise crop yields. Respondents in the case studies and formal sur- 
veys consistently reported significant increases in yields from the use of SFR practices. 
This is consistent with farmer-managed trial data. 

• Despite being used by a number of poor households and having an impact on yields, 
SFR's impact at the household level is modest. This is due to the small land sizes under 
SFR and because the weak rural economy is not conducive to investment and develop- 
ment. As a result, technological innovations alone are likely to have limited short-term 
impact. Poverty alleviation should encompass other sectors as well. 

Strategies for Addressing Poverty and Soil Fertility 

SFR technology interventions imply assumptions about the role of agriculture in people's 
livelihoods that may not be true. The assumption that poverty can be reduced through farm- 
ing is not necessarily reflected in the investments in livelihood activities made by people in 
the region. In fact, their decisions are embedded in their economic circumstances, cultural and 
normative frameworks, and social identities. 

Identifying agricultural strategies to reduce poverty is difficult due to low prices, variable cli- 
mate, and the high cost of profitable investment. Small land holdings, in turn, limit the amount 
of diversification that households are willing to undertake. This study shows that even when 
progress is made, households can easily slip back into poverty. Therefore, in addition to the 
generation of production and income, the need exists for insurance through investment in risk- 
buffering assets. 

The soil fertility systems being disseminated are useful options for farmers, and many 
farmers who have never before invested in soil are giving them a try. There are clear limi- 
tations to the use of improved fallows and biomass transfer, however. Small farm sizes, for 
example, limit the extent to which niches can be found to produce the green manures. The 
technologies are therefore best perceived as components of a larger farm-level integrated soil 
fertility management strategy. Consequently, dissemination strategies should encompass a 
range of management practices for addressing the problem of poor soil fertility. 



Background to the Study 

This study of soil fertility replenishment (SFR) technologies in western Kenya is one of 
seven case studies that comprise a research project examining the impact of agricultural 
research on poverty. The impetus behind these studies is the belief that "the contribution 
of CGIAR and national agricultural research centers to food production is well established. 
However, the extent to which the poor have benefited from agricultural research is less certain" 
(IFPRI 2000). To address this concern, CGIAR's Special Program for Impact Assessment 
(SPIA), formerly the Impact Assessment and Evaluation Group (IAEG), requested that IFPRI 
develop and coordinate a project to strengthen capacity for poverty assessments, not only to 
identify the conditions under which agricultural research is a sound investment for reducing 
poverty, but also to improve the targeting of research priorities to the changing needs of the 
poor. Each study aimed to develop methods for evaluating the impact of agricultural research 
on poverty in the context of different agricultural technologies and within different country, 
social, and institutional settings. It is hoped that collectively the studies will point toward a 
conceptual framework that agricultural research centers can draw upon for impact assessment 
work, and that will also help them to identify research priorities and guide technology design 
to increase the impacts on poverty in the future (IFPRI 2000). 

The project started with a review and synthesis of the literature on the relationship between 
agricultural research and poverty, and this stage was completed in 1999. The second phase, 
begun in September 2000, involved the seven empirical case studies. The seven studies are 
identified in Table 1.1. Each case study focuses on a set of research questions driven by the 
nature of the technology under study and its context. Five of the case studies (the exceptions 
are the China and India studies, which used econometric analysis of secondary data and did 
not involve new fieldwork or mixed research methods) address a set of common themes, and 
used "livelihoods conceptual framework" (see Chapter 2) as a starting point for the research 
design. Each of the five studies involves a combination of previous and new quantitative data 
sets and qualitative methods. All seven studies have been synthesized into a set of findings and 
recommendations on future impact assessment work (Meinzen-Dick et al. 2004). 

Introduction to the Western Kenya Case Study 

Western Kenya is an area of high poverty and low agricultural productivity, especially when 
contrasted with its potential. The International Center for Research in Agroforestry (ICRAF) 
was invited by the government of Kenya to base a research program at the Maseno Research 
Centre on the Kisumu-Busia road. The first activity was a diagnostic study of farmer condi- 
tions, problems, and opportunities in the area. One of the problems highlighted by farmers was 


Table 1.1 Phase 1, Wave 1 case studies of impact of agricultural research under the 
IFPRI/SPIA project 



Lead CGIAR center 




Modern rice varieties 
Polyculture fishponds 
Improved vegetables 

Modern rice varieties 

Soil fertility management 

Modern maize varieties 

Creolized maize varieties 

International Rice Research Institute (IRRI) 

International Food Policy Research Institute (IFPRI) 

International Center for Research in Agroforestry (ICRAF) 


Centra Internacional de Mejoramiento de Maiz y Trigo 

Agricultural research investments IFPRI 
Agricultural research investments IFPRI 

poor soil fertility. During the same period, 
other pioneering work from Wageningen 
(Stoorvogel and Smaling 1990) established 
that nutrient outputs from western Kenyan 
farmers' fields exceeded inputs by a wide 
margin. Drawing from this evidence, 
ICRAF established a research program to 
address soil fertility problems in western 
Kenya. The rationale for why such a pro- 
gram was expected to alleviate poverty is 
discussed in Chapter 4. 

ICRAF tried out the improved fallow 
technology in western Kenya in 1991, both 
under experimental circumstances and on 
farms. At that time the only species used 
was Sesbania sesban, an indigenous species 
that had proven its potential in southern 
Africa (Kwesiga and Coe 1994) and was 
a prolific biomass producer under western 
Kenyan conditions (Onim, Otieno, and 
Dzowela 1990). The agronomic perform- 
ance and economic profitability of Sesbania 
fallows were studied in detail (Hartemink et 
al. 1996; Swinkels et al. 1997; Jama, Buresh, 
and Place 1998a). At that time, alley farming 

was also being tested and a major review of 
that research raised questions as to its per- 
formance and viability in western Kenya. 
Thus, there was a period of stagnation 
(1994-95) during which there was little dis- 
semination of soil fertility technologies to 

In 1996, new fallow species had been 
introduced with promising results, and the 
directors of ICRAF, the Kenya Agricultural 
Research Institute (KARI), and the Kenya 
Forestry Research Institute (KEFRI) decided 
to intensify efforts in research and dissemi- 
nation of improved fallows. This was cat- 
alyzed by recent success in Zambia, where 
yields and profits were found to increase 
substantially from improved fallows, com- 
pared to low-input farmer practices. ' More- 
over, a study of existing soil fertility manage- 
ment practices in western Kenya revealed 
that about 50 percent of farmers did leave 
some of their fields uncultivated for at least 
a season on a regular basis (De Wolf, Rom- 
melse, and Pisanelli 2000). Screening trials 
resulted in the selection of new species that 

'Many farmers had been using fertilizer in Zambia, but the government subsidization of fertilizer price and credit 
halted after structural adjustment policies were adopted. 


in most cases were shrubs and had a shorter 
life cycle than Sesbania sesban. The most 
promising and widely used species are Cro- 
talaria grahamiana and Tephrosia vogelii 
(Niang et al. 1999). Other aspects and 
management options that were tested under 
research conditions are planting densities 
(Niang et al. 1999), the addition of inorganic 
phosphorus fertilizer (Jama, Swinkels, and 
Buresh 1997; Jama, Buresh, and Place 
1998a), the effect on weeds (Niang et al. 
1996), effect on nematodes (Desaeger and 
Rao 1999), and minimum-tillage planting. 

Also from the mid-1990s, testing began 
of local shrubs in collaboration with the 
Tropical Soils Biology and Fertility Pro- 
gramme (TSBF) to look at the potential to 
supply nutrients to maize crops. One species, 
tithonia, was found to be the most promis- 
ing among several because of its ease of 
establishment, easy handling (some other 
species have thorns or sharp leaves), high 
concentration of nitrogen, and good yield 
impacts on crops. In the beginning, tithonia 
leaves were gathered from roadsides or farm 
boundaries and applied to plots at planting 
time. Later, farmers explored a whole range 
of management options, but in all cases, a 
system of biomass transfer was practiced 
(growing the shrubs in one place and apply- 
ing the biomass in another). 

From 1996 to 2001, extensive on-farm 
experiments were conducted to assess the 
potential of fallows and biomass transfer 
using these species, often in combination 
with phosphorus fertilization. Agronomic and 
economic performance was studied within 
these trials. Initial efforts at the beginning of 
1997 were focused in a pilot project area in- 
volving 17 villages distributed mainly in the 
neighboring districts of Vihiga and Siaya. 
Village committees were established to help 
facilitate information flows between the 
community and research staff. In addition, 
field technicians were made available to 
many of the villages for a period of about 
two years. Wide-scale dissemination of im- 
proved fallows and biomass transfer across 
western Kenya started at the end of 1998. 

To disseminate the technology more 
widely, the research partners developed 
partnerships with the Ministry of Agricul- 
ture and with nongovernmental organi- 
zations (NGOs) such as CARE-Kenya, 
KWAP-Busia, Hortiquip-Vihiga, SCODP- 
Siaya, NCCK-Kisumu, VI-Agroforestry 
project-Kitale, IRAM-Vihiga, and many 
community- based organizations. Some inter- 
action with these partners took place in ear- 
lier phases of the technology development 
process, but these intensified in 1998. The 
NGO partners integrated agroforestry op- 
tions into their existing portfolios of options 
for communities and disseminated them 
using existing approaches, including training 
of primary contact farmers, field days, and 
exchange tours. To make this happen, the re- 
search partners trained extension and devel- 
opment organization staff on the establish- 
ment and management of the agroforestry 
technologies and provided them germplasm 
of species new to the area. Many field days 
were conducted, first at researcher-managed 
sites and later at farmers' fields. Finally, ex- 
tension materials were developed for use by 
development agents. 

In several instances, organizations, com- 
munities, or individual farmers took the 
initiative to seek information. Almost every 
day, the research center at Maseno received 
visitors requesting germplasm or information 
about the agroforestry systems. One particu- 
larly interesting case was West Kanyaluo, 
located in Rachuonyo District (one of our 
case study sites for the quantitative work). A 
subchief (leader of a sub-location) heard 
about the fallows and led a small group of 
farmers to Maseno. Having received some 
seed and information, a few farmers planted 
the fallows. The following year, the subchief 
returned to acquire additional seed for the 
community. More than 100 farmers are now 
practicing improved fallows in the commu- 
nity without ever having received technical 
assistance from research or extension. 

This study of SFR technology was one of 
a set of CGIAR case studies examining the 
impact of agricultural research on poverty. 


This particular ICRAF technology was se- 
lected for this study because it was an ex- 
ample of natural resource management re- 
search as opposed to the more common crop 
variety research. Because the agro forestry 
technologies offered an affordable option 
for soil fertility improvement, it was ex- 
pected that rates of use and adoption would 
be relatively high among the poor. Further, 
there was some question whether the non- 
poor would perceive any benefits of agro- 
forestry compared to fertilizer. However, 
whether the poor can substantially benefit 
from agroforestry is still an empirical issue. 
It may depend on their understanding of 
how to effectively manage the systems as 
well as their capacity and willingness to 
increase their land and labor investment in 
these systems. 

The study was also unique in its focus 
on comparing approaches to dissemination 
of the technology. Exploring dissemination 
processes speaks to debates around social 
capital, empowerment, and participatory de- 
velopment. Technology is mediated by so- 
cial processes and social relationships. In 
addition to examining how these processes 
unfold, the study explores several hypothe- 
ses related to the use of local organizations 
and other forms of participation for dissem- 
ination: that social capital will be enhanced; 
that social divisions will emerge; that farm- 
ers will be newly empowered in certain di- 
mensions; and that existing power relation- 
ships will be reinforced. Although these 
may appear contradictory, we found that 
they occur simultaneously, with a range of 
effects on different individuals. Also, in dis- 
aggregating the focus groups into women 
and men, and poor and less poor farmers, a 
hypothesis suggesting socially differentiated 
impacts is implicit. The findings have im- 
plications for policy and program choices 
related to forms of farmer participation in 
technology development and dissemination, 
and suggest the importance of understand- 
ing local social dynamics in designing pro- 
gram interventions. 

It should be mentioned that the range 
of issues covered in the study required the 
use of mixed research methods and interdis- 
ciplinary perspectives. Issues pertaining to 
local perceptions of poverty, the mediation 
of technology by social processes, and the 
role of gender, power, and other social con- 
structs in understanding adoption and impact 
could be explored only by using qualitative 
research methods and sociological perspec- 
tives. These were combined with quantita- 
tive measures of adoption and impact, and 
economic analysis. The quantitative analyses 
proved valuable in identifying the preva- 
lence of patterns of adoption and impact 
relationships among the general population 
and the poor. We also draw on long-term 
knowledge of the region, based on work by 
members of our study team and others. The 
study was designed using a livelihood con- 
ceptual framework as a starting point, draw- 
ing on concepts of vulnerability, access to 
and limitations on combinations of assets 
(e.g., natural, human, and social capital), 
and the importance of institutions and pro- 
cesses. However, other constructs from eco- 
nomics and sociology were introduced as 
required. This was done not so much to pro- 
mote a particular, alternative paradigm, but 
rather in the spirit of handling the impor- 
tant research questions raised by the entire 
team, comprising individuals of diverse 
backgrounds and experiences. 

Outline of the Report 

Chapter 2 presents the methods, with an 
analysis of how they were used to address 
the many detailed questions that guided the 
research. It also reports on the sampling 
procedures and outcomes. Chapter 3 pro- 
vides a contextual background to the study at 
the national Kenya level, the western Kenya 
region, and also by ethnicity, as we con- 
ducted studies in both the Luo and Luhya 
communities. Included, too, are the assump- 
tions of the soil fertility replenishment re- 
search program. Chapter 4 probes the con- 


cept of poverty from official, researcher, and 
local perceptions and definitions and in- 
cludes some analysis of our data. Chapter 5 
presents data on livelihood strategies. While 
largely descriptive, it also offers insights as 
to the compatibility between investments in 
soil fertility and the pursuit of better liveli- 
hoods. Chapter 6 focuses on adoption of soil 
fertility practices, describing the process in 
the pilot and non-pilot villages. It proceeds 
to explore in some depth the patterns of 
adoption across different types of house- 
holds, including the poor versus the non- 
poor. Chapter 7 then explores in great detail 
the extent to which various productivity and 
welfare impacts may have taken place as a 

result of adoption of soil fertility replenish- 
ment practices. Chapters 8 and 9 address the 
effect that various dissemination organiza- 
tions and their methods have on knowledge 
acquisition, adoption of the technologies, 
and the communities in which they are intro- 
duced. Chapter 8 focuses on farmers' evalu- 
ations of different dissemination methods 
introduced by external organizations, while 
Chapter 9 focuses more on local dissemina- 
tion processes, impacts on social and human 
capital, and other aspects of social relation- 
ships. Last, Chapter 10 includes a summary 
of methodological and empirical findings, 
along with considerations for future poverty 
alleviation programs in western Kenya. 


Research Methods 

The Livelihoods Framework 

A livelihoods framework was used in the research for this report to understand and link 
issues of agrotechnology development with social processes and poverty. It was used 
first as a guide to collect data and to phrase questions for the case studies. At a later 
stage, it guided the interpretation of data and structured the report. Toward the end of the 
report, we will assess the framework. 

The livelihoods framework is a tool to improve our understanding of livelihoods, particu- 
larly those of the poor. "Livelihood" has been coined as an umbrella concept for research as 
well as for development planning. It involves a framework of analysis that has two main ob- 
jectives. First, it links holistically the variety of ways by which rural people manage to make 
a living for themselves within the contexts in which they operate. Second, it attends to the 
processes that shape these endeavors, and to the activities of institutions and individuals that 
are external to the communities under consideration, but intervene in the way people try to 
make a living. 

The idea behind this framework can be summarized in the definition formulated by Robert 
Chambers and Gordon Conway in the mid-1980s, which stated that livelihood comprises the 
capabilities, assets (including both material and social resources), and activities required for a 
means of living. A livelihood is sustainable when it can cope with and recover from stresses 
and shock and maintain or enhance its capabilities and assets both now and in the future, while 
not undermining the natural resource base (Carney 1998, 4). 

Such a definition links livelihoods with people's resources, capabilities, and activities, that 
is, what people do with such resources or assets. We shall return to a discussion of resources 

The livelihood framework serves a variety of purposes and interests. Approaches by policy- 
makers that provide a blueprint or preconceived plan for improving the conditions of living 
for poor countries and poor people have been widely criticized. For policymakers, and partic- 
ularly for the international donor community, "livelihood" provides a framework that focuses 
on poverty within the contexts of the people who are poor, and on the processes that underlie 
poverty. For consultants who operate in the field of development, "livelihood" represents a 
framework for the formulation of development projects that focus on the people being affected 
by the project and the variety of ways in which they might be affected. For social scientists, 
such as anthropologists, sociologists, and economists, "livelihood" provides a framework for 
a holistic interpretation of the dynamics of development and the different rhythms of change. 
For plant breeders, soil scientists, and other technologists, the livelihood framework links their 


specific work and capacities with what peo- 
ple are capable of doing, what they are look- 
ing for, and how they perceive their needs. 

The livelihood framework thus provides 
a guide for research and intervention. "Live- 
lihood" focuses on the fact that the people di- 
rectly affected by poverty, through attempts 
to alleviate it, are striving to make a living, 
preferably above the level of mere survival. 
In doing so, they try to create and embrace 
new opportunities, such as trade and crafts, 
food processing, new technological innova- 
tions for agricultural production, and labor 
migration. At the same time they may have 
to cope with risks and uncertainties, such as 
erratic rainfall, diminishing resources, pres- 
sure on the land, changing life cycles and 
kinship networks, epidemics such as human 
immunodeficiency virus/acquired immuno- 
deficiency syndrome (HIV/ AIDS) and other 
diseases, chaotic markets, increasing food 
prices, inflation, and national and inter- 
national competition. These uncertainties, to- 
gether with new emerging opportunities, im- 
pinge on how material and social resources 
are managed and used, and on the choices 
people make between different sets of values 
and identities associated with such usage. 

Livelihoods include activities both within 
the locality and stretching beyond it. These 
activities concern, for example, agricultural 
practices, production of knowledge and 
gathering information, trading, migratory 
labor, transport, the search for money via 
credit schemes, or the negotiation of sex. 
Wide-ranging interpersonal networks link 
rural and urban areas, on-farm work and 
off-farm work, and dryland farming and 
irrigated farming. Livelihood by definition 
transcends the boundaries between eco- 
nomic sectors (agriculture vs. industry, for- 
mal employment vs. informal activities). 
Livelihoods often transcend geographical 
boundaries, particularly those between urban 
and rural environments. In other words, peo- 
ple do not live and work only in domains 
where the boundaries are defined by bureau- 
cracies, like rural and urban or districts and 

land resettlement schemes, or by natural con- 
ditions, such as watersheds or agroclimatic 
zones (Rhoades 1998). People's livelihoods 
are analytically situated and practiced by 
people in social spaces with boundaries de- 
fined by social networks, relationships, and 
identities. These spaces are fluid, constantly 
changing, and are shaped and constantly re- 
negotiated by people themselves. They are 
often understood as "arenas" since they 
are subject to struggles and negotiations. 
More often than not, such arenas also involve 
the contestation of knowledge — clashes be- 
tween different bodies of knowledge. 

Livelihoods, then, can be understood 
only by mapping out the various actors 
(farmers, their families, administrators, 
traders, extension workers, state institutions, 
and so on), and the networks and social 
relationships between them. These actors 
pursue a variety of identities, interests, and 
needs — shaping, in turn, the particular 
strategies they devise to improve their con- 
ditions of living and their well-being. These 
strategies are invariably multiple, implying 
both that there are a variety of ways to sus- 
tain a livelihood and that people undertake 
manifold activities to obtain food, shelter, 
money, and identity. Some individual and 
corporate interests collide, as do their strate- 
gies and discourses of development. For in- 
stance, development interventions by exter- 
nal agencies, such as state institutions, are 
often shaped by ideologies of modernization, 
resulting in a lack of attention to local knowl- 
edge, cultural repertoires, and practices. 

The U.K. Department for International 
Development (DFID) developed the sustain- 
able livelihoods (SL) framework, together 
with social scientists, to both capture liveli- 
hoods and to analyze how livelihoods change 
over time (see Figure 2. 1). The arrows within 
the framework are used as shorthand to de- 
note a variety of relationships, all of which 
are highly dynamic. None of the arrows 
implies direct causality, although all imply 
a certain level of influence. Although the 
previous discussion has referred to these 


Figure 2.1 Sustainable livelihoods framework 

F = Financial capita] 
H = Human capital 
N = Natural capital 

P = Physical capital 
S = Social capital 




and access 





Levels of 



More income 

Increased well- 


food security 

use of natural 
resource base 

components, we now focus on a few core 
concepts: the importance of context, trans- 
forming processes, and livelihood assets or 


Context is fundamental to understanding 
livelihoods and is pervasive in this report. 
Context refers not only to broad political 
and economic structures, but also to the im- 
mediate physical, social, and cultural envi- 
ronments. Contexts vary enormously, as do 
development processes. These processes are 
locally specific, shaped by history, cultural 
repertoires, economic and political relation- 
ships, and the natural environment. Live- 
lihood is essentially contextual: livelihoods 
can be captured and understood only in par- 
ticular contexts. 

For those with few resources, these 
contexts are often risky, making people 
vulnerable to shocks, stresses, and changes. 
Changes in growing populations, family 
composition, governance, technology, health 
and diseases, as well as changes due to con- 
flict, seasonal variation, drought, and pests, 
affect what people do and may enlarge or 
limit their room for maneuver. Specific pro- 

gram interventions attempt to reduce some 
aspects of vulnerability. 

The context of this particular research in 
western Kenya is one of endemic poverty; 
declining soil fertility due to land degrada- 
tion; failing markets, particularly for agri- 
cultural inputs (that is, seeds and fertilizer); 
low farmgate prices for cash crops such as 
maize and coffee; lack of alternative employ- 
ment opportunities; low incomes; increas- 
ing food prices; and reduced landholding 
as a result of population pressure and con- 
tinuous subdivision of land. Cash crops in- 
troduced earlier in the region, such as cof- 
fee, sugarcane, and cotton, have so far not 
proved sustainable. Their acreage has grad- 
ually declined, leaving only signs of their 
previous existence. Some farmers, however, 
still maintain a few coffee bushes in the 
hope that there will one day be a market for 
coffee. Those who grow sugarcane do not 
earn much because of monopsonistic mar- 
ket conditions whereby the processors capi- 
talize on the imperfect market conditions. 

Many "households" are currently net 
food buyers rather than sellers. In the past 
they relied on the flow of cash from migrants 
for the purchase of food, although today this 


rural-urban connection is jeopardized be- 
cause decent off-farm employment opportu- 
nities in urban centers are not easy to find 
(see also Omosa 1998). For those unable to 
afford inputs, soils may decline to the extent 
that the farmers then grow what are mainly 
known as "poor man's crops," such as sor- 
ghum, cassava, sweet potatoes, and millet. 
They may also move away from hybrid 
maize, turning back to local maize varieties. 
However, those who cannot afford inputs 
may also experiment with alternative ways 
of reproducing soil fertility, including agro- 

One "structural" feature of agriculture 
in the region is the predominance of non- 
commodity relationships and the role of a 
seemingly localized market. Farmers' strate- 
gies are clearly geared toward reducing cash 
outlays for agriculture, searching for low- 
cost resources and distancing them from 
markets. In such a context, from a theoreti- 
cal point of view, agroforestry-based tech- 
nologies would fit perfectly. 

This context also has clear political, in- 
stitutional, and economic dimensions. Ex- 
tension services are declining as a result of 
natural attrition as well as retrenchments 
following the implementation of the Struc- 
tural Adjustment Programme by the World 
Bank in the early 1990s. The situation is 
now even worse, since the World Bank 
stopped funding the agricultural develop- 
ment sector in early 1997. Extension depart- 
ments have few funds and their officers are 
virtually grounded. 2 Furthermore, input 
prices have increased while farmgate prices 
for outputs can hardly keep up with infla- 
tion. Under such conditions it is difficult to 
sell an extension message based on com- 
moditized inputs. Such messages become 
increasingly irrelevant, even to extension 

An important part of the economic di- 
mension is the poor state of the country's 
economy. In the formal sector, more jobs 
disappear than are created because of lack 
of investment (e.g., there has recently been 
a net outflow of foreign direct investment), 
and the government is trimming staff to meet 
the conditions for a restoration of World 
Bank and International Monetary Fund 
(IMF) borrowing. In these declining eco- 
nomic conditions and lack of alternative 
employment, people return home and start 
farming again. 

Transforming Processes 

The livelihood framework specifies in more 
detail the transforming processes and struc- 
tures that change and shape the contexts 
in which people try to make a living. The 
transforming processes and structures thus, 
directly or indirectly, impinge on people's 
livelihoods (Carney 1998; Scoones 1998; 
Ellis 2000). They operate at various levels. 
The social level includes changing relation- 
ships and the structures of kinship, gender, 
and age. There is a cultural level incorpo- 
rating customs, religion, and other beliefs, 
including notions of development. The 
economic level involves investments, global 
and local markets, prices, international com- 
petition, and technologies. The political level 
covers governance and policies, tribal au- 
thority, the state, and wars and conflicts. It 
is linked to the judicial level, which covers 
laws of the state, customary laws, and such 
things as land tenure and rights. The natural 
environment is also relevant, particularly 
when natural calamities occur or when the 
land erodes away. The framework thus aims 
to link the micro and macro levels of devel- 
opment. It attends to how these interact with 
each other, and how changes at any level 
transform what is happening and what is 

2 In 2001, a new extension program funded by the Swedish International Development Cooperation Agency (Sida) 
was launched. This will provide resources to a team of extension agents who will then concentrate in one cluster 
of villages per division in all of the medium-to-high potential areas in Kenya. 


possible at the micro level. This also brings 
in the necessary dynamic element of change: 
processes of transformation that shape live- 
lihood contexts. 

Luhya and Luo agriculture generally is 
subject to transformations at many levels. 
The social level constitutes the transforma- 
tions at the level of labor relations where the 
ability or inability to mobilize labor is an 
important issue for agricultural production 
(despite the perception of the outside ob- 
server that labor is widely available). More- 
over, labor is drawn away from agricultural 
production to other economic activities on 
and off the farm. The impact of HIV/AIDS 
on labor relations is also notable. 

Land tenure relationships, particularly 
the inheritance of land, are subject to con- 
flicts owing to a confusing mix of customary 
and state law; the latter fosters private land 
ownership. 3 Furthermore, land and labor 
relationships are obscured by jealousy. Jeal- 
ousy is embedded partly in interpersonal 
relationships, for example, between brothers, 
and partly in the kinship and lineage com- 
plex. The latter seems to be more important 
for the Luo than for the Luhya. 

The natural environment of Luo and 
Luhya agriculture is one characterized by 
bimodal rainfall patterns providing the op- 
portunity for planting twice a year. The 
highland ecology offers considerable scope 
for tree growing and is characterized by 
genetic variation. For instance, various 
so-called local maize varieties are available, 
providing farmers with opportunities to 
breed and select seeds themselves, rather 
than relying on hybrid maize seed. Occa- 
sionally, however, the rains fail, causing 
people to plant late. 

Soil fertility is among the major ob- 
stacles to increased agricultural production. 
Nitrogen amendments can be increased ei- 
ther through purchased fertilizers or through 

on-farm organic techniques such as animal 
manure or agroforestry. The shortage of 
phosphorus, however, can be adequately ad- 
dressed only through buying the nutrient. 

As referred to earlier, markets generally 
are imperfect, in terms of access, availabil- 
ity, and price setting (market structure and 
conduct) as well as an underdeveloped in- 
frastructure (marketing context). The recent 
liberalization of markets for inputs (seeds, 
fertilizers, output) made these markets 
rather chaotic and unreliable. Cash returns 
to farm labor through involvement in such 
markets is not as high as farmers would 
hope. Such uncertainties and limitations 
affect peoples' social identities as well as 
incomes. Potential farmers are left asking 
themselves whether they can make a living 
based on agriculture. 

Resources, Capitals, Assets 

Livelihood resources are often categorized as 
the vital "capitals" that one needs to achieve 
a sustainable livelihood. These resources, as 
the framework specifies, include: 

• Human capital such as labor, skills, 
knowledge, creativity, experience, drive 
toward experimentation 

• Natural capital in the form of natural 
resources such as land, water, minerals, 
crops, forest, and pastures 

• Physical capital that can be food 
stocks, livestock, tools, and machinery 

• Financial capital in the form of money, 
loans, or credit; state transfers; remit- 
tances; savings 

• Social capital, which concerns the 
quality of relationships among people 
and the extent to which one can count 
on support by the family or mutual 

The coining of "social capital" highlights 
the idea that livelihoods are seldom con- 

3 See Hebinck and Mango (2001) for an analysis of land conflicts in the Luo region. 


structed on an individual basis only, but 
rather are embedded in interpersonal net- 
works. Three core elements constitute social 

• Relationships of trust, reciprocity, and 
exchange between individuals, often 
embedded in specific local forms of 
organization and shaped by cultural 

• Connectedness, networks, and groups, 
including access to wider institutions 
and so-called "distant" actors 

• Rules, norms, and sanctions that are 
often, but not always, mutually agreed 

These resources or assets will be described 
in detail in Chapter 5 with a focus on under- 
standing rural livelihoods in the region. 

Research Questions 

The development of the research ques- 
tions was a three-stage, iterative process. 
The first step, outlined in the introduction, 
involved the development of a research 
proposal that outlined six "researchable" 
questions that spoke to specific objectives 
of the wider SPIA/IAEG work within an SL 

The second stage was a workshop, held 
in Maseno and Nairobi in December 2000, 
to obtain stakeholder input into research 
issues and methods, important elements of 
vulnerability contexts, assets and trans- 
forming structures and processes, and rele- 
vant findings from previous and ongoing 
work. Stakeholders included academics 
from the fields of development studies and 
agricultural economics, ICRAF staff, farm- 
ers, chiefs, NGO staff working in the re- 
gion, and teachers. Based on their input and 
subsequent discussion, the original proposal 
was revisited to determine whether the 
original questions remained appropriate; 
identify relevant subthemes and research 
questions; clarify sources of existing in- 

formation; and identify where additional 
knowledge should be gained. 

After discussion at the stakeholder 
workshop and among the research team, it 
was decided to retain the existing research 
questions, with some modifications to lan- 
guage. However, in response to presenta- 
tions on the overall study objectives, the 
SL framework and the outline of the Kenya 
case study (as presented in the IAEG/SPIA 
research proposal), a number of useful sug- 
gestions were provided on the development 
of appropriate subthemes and more detailed 
research questions to address the larger re- 
search issues. These included (parentheses 
refer to the place within the SL framework 
that these suggestions fell): 

1 . The importance of understanding the 
social context of farming/agriculture in 
the region (transforming structures and 

2. How poverty is understood locally 

3. The tensions between agriculture and 
off-farm work (livelihood strategies) 

4. Farmer adaptation of the "standard" AF 
practices (vulnerability contexts; trans- 
forming structures and processes; 
livelihood strategies; and livelihood 

5. Distinguishing between adoption/non- 
adoption and the intensity of adoption 
as defined by the area planted and kind 
of AF technology, for example, im- 
proved fallow; biomass transfer (trans- 
forming structures and processes; 
livelihood strategies) 

6. The importance of disaggregating by 
gender (vulnerability contexts, assets, 
transforming structures and processes; 
livelihood strategies, livelihood 

7. The mechanisms by which crop 
surpluses are translated into increased 
stocks of assets; what constrains 
choices (vulnerability contexts, assets, 
transforming structures and processes; 
livelihood strategies, livelihood 


8. The different dissemination approaches 
in the wider region (transforming struc- 
tures and processes) 

9. Changes in socioeconomic differentia- 
tion in these localities and the extent to 
which they are an outcome of differ- 
ences in the intensity of adoption or 
use of agroforestry 

The third stage was a three-day meeting of 
the principal researchers in Nairobi, in April 
2001, that focused on reviewing the research 
questions and subthemes in light of prelim- 
inary findings. These are outlined in Tables 
2.3 and 2.4. 

Quantitative Methods 

Quantitative methods are used for the fol- 
lowing analyses: (1) characterizing and 
identifying the poor, (2) assessing the abil- 
ity of alternative dissemination approaches 
to reach the poor, (3) assessing the extent to 
which the poor are using the agroforestry 
technologies, and (4) assessing the degree 
to which the poor are benefiting from the 
use of agroforestry technologies. In terms 
of identifying the poor, ICRAF had already 
used a survey instrument containing many 
wealth-related variables (identified through 
wealth-ranking exercises) to make a prelim- 
inary classification of households according 
to wealth. This classification served as a 
stratification for the 1999-2000 baseline 
survey of food and nutrition and as well for 
the case study selection in the qualitative 
analyses (see section "Case Studies on Live- 
lihoods, Adoption, and Impacts"). However, 
this is used as a preliminary sampling tool 
only. In the evaluation of the impact of 
agroforestry on the poor, more rigorous 
definitions of poverty are used, including 
quantitative measures of assets and baseline 
consumption or expenditures. 

For the remainder of the quantitative 
analysis, it is first useful to distinguish the 
analysis for households within the pilot 
project area and those outside. Within the 

pilot project area, an earlier baseline assess- 
ment of assets, expenditures, and food con- 
sumption had been made. The baseline was 
given to 120 households, stratified in the 
following ways: 

• 60 Luo, 60 Luhya 

• 60 not yet using agroforestry, 60 start- 
ing to use agroforestry 

• 40 from the poorest tercile, 40 from 
the middle tercile, and 40 from the 
wealthiest tercile 

• Balanced design so that approximately 
10 in each of 12 cells were enumerated 

The baseline was administered at four differ- 
ent times of the year, and for food consump- 
tion, was measured by 24-hour recall for 
three- and two-day periods. In the follow-up 
for 2002, only one time period was enumer- 
ated, that being the April-May period just 
before the harvest from the long rains. It is 
during this period that differences in crop 
yields and incomes have the most impact on 
food consumption. 

The analysis focuses on the changes in 
assets, expenditures, and consumption over 
the 1999-2002 period, and how these are 
related to the use of the agroforestry sys- 
tems, conditional on initial wealth status. 
Two-stage methods are used in which the 
choice of use of the agroforestry technolo- 
gies will be modeled in the first stage and 
the impact from the use modeled in the sec- 
ond stage. 

In addition to this detailed quantitative 
study of approximately 120 households, 
ICRAF had been monitoring the use of the 
agroforestry systems between 1997 and 2001 
for about 1,600 pilot project area house- 
holds. Together with a 1997 census instru- 
ment, it was possible to assess dynamic pat- 
terns of use of the technology by the poor 
and less poor. Multinomial logit regressions 
are made to identify factors associated with 
non-adoption, dis-adoption, and adoption. 
We also assess the role of household factors 
in the intensity of adoption both spatially and 
temporally. Within the pilot project area, 


ICRAF and its partners were the main en- 
gines of dissemination, so that there is really 
no variation in dissemination techniques 
within the pilot project area. 

Outside the pilot project area, six vil- 
lages were selected as research sites: Luo: 
Muhande-Arude (Siaya), Ugunja (Siaya), 
and West Kanyaluo (Rachuonyo); and 
Luhya: Mwitubi (Vihiga), Shinyalu (Ka- 
kamega), and Bukhalalire (Busia). The sites 
were selected because they were among the 
earlier places where dissemination occurred 
and they were reached using different dis- 
semination methods. CARE had operated in 
Muhande and was the principal agent there, 
until they left in 1999. An umbrella NGO 
operated in Bukhalalire, a local catchment 
committee (formed by extension to be their 
liaison in the community), and a local CBO 
in Ugunja. In Shinyalu, KARI had estab- 
lished a participatory learning and action 
approach with the community. Finally, in 
West Kanyaluo, a local assistant chief 
learned of the technologies and spearheaded 
collaboration with ICRAF by funding a trip 
to Maseno. He and other farmers visited 
about three times in all to obtain informa- 
tion and seed. 

In each village, 60 households were 
selected, approximately half of which were 
found to be using agroforestry. 4 The survey 
instrument was broad, covering basic house- 
hold characteristics, acquisition of informa- 
tion on agroforestry, use of agroforestry, and 
impacts from agroforestry. In addition to 
agroforestry, identical questions on infor- 
mation and use were asked about other soil 
fertility management strategies. No baseline 
survey was conducted for these households, 
so in only a few cases could data on changes 
be measured, for example, for physical as- 
sets. In other cases, impacts are assessed 
only based on current (2002) values. 

Similar to the analysis of the pilot 
project households, regression techniques 
will be used to assess the extent to which 
the poor are receiving information about 
agroforestry (conditional on dissemination 
method, too), the extent to which the poor 
are using agroforestry (and comparing these 
rates to other soil fertility practices), and 
finally assessing the impacts of agroforestry 
on a variety of variables including crop 
yields, soil fertility, assets, and other wealth 

Qualitative Methods 

To obtain participant-defined characteriza- 
tions of livelihood strategies and outcomes, 
it was decided to aim for a series of case 
studies. It was also decided that the field- 
workers would try as much as possible not 
to identify themselves with ICRAF (and 
thus avoided being dropped off by ICRAF 
vehicles). This was to prevent participants 
from giving answers they felt ICRAF would 
want to hear. 

The intrinsic value of this work is that it 
can be used to direct or extend some of the 
quantitative investigations. For example, it 
can help specify appropriate livelihood out- 
comes for further study, patterns, and rea- 
sons for adoption/adaptation/dis- and non- 
adoption; pick up stories about conflicts 
with implementing agencies and their impli- 
cations for the future; and so on. In addition 
to providing directions for the quantitative 
research, the qualitative research can also 
provide insights and explanations that the 
survey data cannot. The qualitative research 
was composed of two major data collection 
exercises, mirroring the two sets of central 
research questions. The first concerns live- 
lihoods, adoption, and impacts; the second 
relates to technology dissemination. 

initially, we had intended to stratify on the basis of wealth rather than use of agroforestry. However, the rates of 
use were highly variable and in our second intended village (Central Gem), we found only five farmers using 
agroforestry. Thus, we abandoned the more ideal but risky strategy of stratifying by wealth alone, and opted to 
ensure sufficient numbers of adopters and non-adopters. 


Case Studies on Livelihoods, 
Adoption, and Impacts 

The study draws on a mix of approaches. 
The core method would be a series of 
household-level case studies, which involve 
extensive informal interviews held within 
and outside the household over a period of 
six months, supplemented by participant 
observation. Focus groups would be used to 
confirm findings of individual case studies, 
reconcile divergent findings, and allow a 
wider range of voices to be heard. Such an 
approach, it was thought, would result in the 
wider IAEG/SPIA study containing a range 
of experiences — including both those that 
employ extensive use of group methods (and 
little or no case-study work), and those in 
which case studies took the lead supported 
by group-based activities. 

Acknowledged concerns of undertaking 
lengthy case studies were (1) the replica- 
bility, particularly for future impact assess- 
ment work of studies involving six months 
of fieldwork, and (2) that fewer households 
would ultimately be represented in the 
study. (The follow-up group work, however, 
helps counter the latter disadvantage). In the 
end, it was decided that the depth of under- 
standing gained compared to a more rapid 
assessment approach was substantial and 
thus should be considered as a potential 
method for future impact assessment work. 
One persuasive argument is the fact that if 
research centers are able to spend an ex- 
tended period of time conducting survey 
work, then an extended period can and 
should be given to qualitative work as well, 
if that work is to provide insights that could 
raise the likelihood of increasing impact. In 
addition to these methods, key informant 
interviews would be used to discuss issues 
with individuals with specialized knowledge. 

In sum, the decision to base much of this 
work on case-study narratives was moti- 
vated by the recognition of the importance 
of going beyond quantitative information, 
to understand perceptions regarding soil 
fertility and its management and reasons 
underlying those perceptions. There is also 

Table 2.1 Case study villages 

Case studies Pilot area Non-pilot area 

Luo land 
Luhya land 



a clear need to go "inside the household" or 
even "beyond the household," partly be- 
cause gender and age may differentiate such 
perceptions and partly because access and 
control of resources may vary within house- 
hold units. 

As in the original research proposal, the 
study focused on households in pilot and 
non-pilot areas (Table 2.1). This is based on 
the idea that being exposed to agroforestry 
programs (e.g., technical support) has an 
impact on the adoption process of such 
technologies. Villages in the non-pilot areas 
also experience different dissemination ap- 
proaches that, in turn, may affect processes 
of adoption. 

Within the villages, criteria such as 
wealth and adopters or non-adopters were 
applied to select households for further qual- 
itative research. Within the wealth criteria or 
perceived relative status of poverty (taken 
from earlier wealth-ranking exercises done 
by ICRAF) and adoption/non-adoption, the 
selection procedure also captured as much 
as possible variations across: 

• Female-headed, male-headed, child- 
headed households 

• Relative importance of agriculture in 
family income (kind, cash) 

• Age groups: "young" and "older" 

• Monogamous and polygamous 

The selection procedure made use of the 
quantitative database that is available from 
ICRAF. Twenty households were identified, 
but only ten ultimately selected per village. 
Among these households, half were users of 
either biomass or improved fallows. Based 
on ethnographic understanding, it was agreed 
that individuals rather than the compound or 


household should be the unit of analysis for 
the qualitative research. The 1997 ICRAF 
survey attempted to sample nuclear house- 
holds rather than a compound. The com- 
pound may comprise more than one house- 
hold: husband, wive(s), dependent children, 
and married children still at home. How- 
ever, it may also be that changes have taken 
place in family composition, off-farm work, 
adoption of technologies, and so forth. This 
case selection then broadens the scope. The 
individuals are then to be treated as entry 
points to the compound only. 

The notion of household may fit more 
closely in the Luhya community than with 
the Luo, since the Luhya do not have com- 
pounds like the Luo. Among the Luhya, 
when a father dies, land is subdivided be- 
tween the mother and the sons who have 
their own plot; each will have a gate. But 
there is much cooperation between these 
units, for instance, sharing labor, oxen, and 

The objective of the qualitative analysis 
was to contextualize and understand what a 
technology does to a society such as the Luo 
and Luhya. Qualitative research by means 
of case studies allows for an analysis of so- 
cial processes. Case studies are instrumental 
in exploring how technologies create new 
and/or transform existing social relationships 
among villagers (particularly intergenera- 
tional relationships) as well as between vil- 
lagers and interveners. Gender is only one 
component of such a social transformation 
process. Similarly, qualitative studies are 
probably better equipped than quantitative 
studies to disentangle what poverty actually 
is and how to operationalize a notion of 
poverty for policy-oriented studies. Poverty, 
being subjective, needs to be understood as 
local people view it as well as how policy- 
makers try to measure it with such means as 
the poverty line. 

Focus Groups on 
Dissemination Approaches 

Because the overall study has emphasized 
evaluating technology dissemination pro- 

cesses, with four of the original eight re- 
search questions focused on these issues, an 
additional data collection exercise was de- 
signed to focus on dissemination. Although 
the household case studies addressed some 
aspects of dissemination, focus groups were 
chosen as the main method for this part of 
the study. This is because of the need to 
speak to a large number of people in a short 
time and because dissemination activities 
take place largely in groups. A group-based 
research method would thus allow people to 
debate their experience of these collective 
activities. The focus groups combined dis- 
cussions with the use of certain Participa- 
tory Rural Appraisal (PRA) methods. 

Research questions to be answered 
through the study were determined through 
two stakeholder meetings, where repre- 
sentatives from government, NGOs, KARI, 
KEFRI, ICRAF, and some local organiza- 
tions discussed their concerns with the dis- 
semination approaches and the kinds of 
questions they wanted to explore. In addi- 
tion, key informant interviews were con- 
ducted with representatives of the main or- 
ganizations that disseminated technologies 
in the villages selected for the focus groups, 
in order to gain their insights and opera- 
tional interests with regard to the research. 
The questions centered around evaluations 
of the disseminating organizations and their 
teaching and outreach methods, their inter- 
actions with the community, access to in- 
formation and barriers, group methods of 
outreach, school programs, quality of train- 
ing and what farmers have learned, and 
community social relationships, including 
solidarity and conflict. 

Most of the questions were addressed 
through group discussions, in which facili- 
tators posed open-ended questions and par- 
ticipants discussed their answers in conver- 
sation that was recorded. However, three 
different participatory exercises were also 
used, involving (1) mapping of formal and 
informal institutions inside and outside the 
village, and information flows between them; 
(2) scoring of external and local institutions; 


Table 2.2 Villages and dissemination approaches in the dissemination study 

Type of village 

Type of dissemination approach 

Disseminating organizations a 

Luo villages 

Muhanda - Arude b - C 



Luhya villages 
Mwitubi b - C 
Bukhalalire b 

TRACE approach 
Village approach 
Catchment area approach 

Catchment area approach 


Umbrella group approach 





"The main disseminating organization is in bold. The rest joined in after the approach was in place and used it to 
reach farmers. Acronyms not defined earlier: KIT, Royal Dutch Institute for Tropical Agriculture; KWAP, Kenya 
Woodfuel Agroforestry Programme; MoARD, Ministry of Agriculture and Rural Development; TRACE, Train- 
ing of Resource Persons in Agriculture for Community Extension. 
b Also survey village. 
c Also case study village. 

and (3) using "ladders" to represent prog- 
ress in knowledge acquisition in each SFR 
technology. These methods are described 
further in Chapter 8, with visual examples 
provided there. 

The focus groups were conducted in six 
villages. The villages were selected to rep- 
resent a range of different dissemination 
methods by different organizations, although 
there is overlap in the forms of teaching 
and outreach used. Among these six villages, 
three were from the Luo ethnic group (Sauri, 
Muhanda- Arude, and Gongo) and three from 
the Luhya group (Ishikhuyu, Mwitubi, and 
Mutsulio). Twenty-four groups were held in 
total, with four groups per village disaggre- 
gated by sex and by poverty level, using a 
previous categorization of households based 
on survey data and wealth-ranking data. The 
basis for these disaggregations was to ex- 
plore whether perceptions and experiences 
with the dissemination methods were dif- 
ferent depending on whether one was male 
or female, very poor or less poor, and Luo 

or Luhya. The villages selected and dissem- 
ination approaches studied are presented in 
Table 2.2. 

Text data from the focus groups were 
coded using HyperResearch data analysis 
software. The results of the PRA exercises 
were put into an Excel spreadsheet and the 
descriptive statistics reported. Data analysis 
involved triangulation of the focus group 
results with the quantitative data and case 
study findings. 

Prioritizing Research 

The various workshops described earlier 
formulated the priority research questions in 
detail as well as the sub-research questions 
and themes. The latter also served as a guide 
for the fieldworkers in their case study work 
and the focus group discussions on poverty 
(Table 2.3) and dissemination strategies 
(Table 2.4). 


Table 2.3 Research design matrix: Assets, vulnerability, and livelihoods 

Location in SL 
or other key 

Main questions 
in proposal 

Subthemes/research questions to explore 

Livelihood 1 . What is the 

outcomes; ability of SFR to 

targeting; social reduce poverty? 

How successful have the technologies been in improving outcome indicators (assets, [see 

below] food security)? 
How do people define poverty and well-being? 
How successful are they at targeting the poor? 
What are the key groups among the poor? 

What are the key factors that contribute to making people poor and to vulnerability? 
What is needed to reduce poverty and what protects against vulnerability? 
Why have some households or individuals benefited and others not (gender, other social 

How was poverty conceptualized at the early stage of ICRAF's work? 
How have interventions contributed to women's decisionmaking ability, control over resources, 

and negotiating ability (creating space/empowerment) with regard to household and 

community? How has it improved or worsened intra-household relationships? 
What are the indirect affects in terms of asset accumulation, access to markets, human capital 

formation, social capital formation/community social relationships, empowerment? 
Has SFR influenced people's interest in farming? (generational differences); and how do they 

perceive these in relation to other options? 






2. What are the 
livelihood strategies 
that people pursue 
and what are the 
dynamic relationships 
between livelihood 
strategies and how 
do SFR technologies 
relate to these? 

3. What factors have 
influenced adoption? 
What factors and 
processes mediate the 
ability of different 
social groups to take 
up new technologies? 

Look at all of these 
questions in terms of 
gender, class, and 
other forms of 

How has SFR technologies affected social differentiation? Has it contributed to increased or 

decreased inequality? 
How do local priorities compare/conflict with government priorities? Or are compatible with 

government priorities in terms of outcomes? 

What are the main categories of livelihood strategies pursued by different groups? 

Who in the household is contributing to what types of work, and why? (attention to generational 

How do different technologies fit within these livelihood strategies (or not)? 
What are the complementarities with other activities, including seasonal activities? 
What are the conflicts with other activities, including seasonal activities? 
How are the technologies adapted by farmers and why are these adaptations made? 
What are the different types of farmers and farming systems? 
How have these farming systems changed over time? 
Why do people choose to pursue certain livelihood strategies (resources/assets; trends; 

identity /how people see who they are)? 
What are the sources of perceived vulnerability (natural, financial, social, and so on)? 
What are the gender and class and other significant power structures in the villages and 

households (that will influence answers to questions below) 

What assets are necessary for adoption (physical, natural, social, human, financial, political, 

What makes a good farmer? How does SFR technologies relate to this? 
What assets did/do people have? 
Why are these assets available or lacking? 
What are the key relationships between different types of assets needed and sequencing? Is 

there scope for substitution among assets? Which are more important than others? 
What is the role of social relationships/social capital in facilitating or constraining adoption? 
Why do some people not take up the technologies, stop using them or reduce their use (broad, 

including attitudes)? 
Who makes the decision to adopt? Who has had input into this decision? What was the basis of 

the decision? 
What cultural and social aspects mediate adoption? 

(continued ) 


Table 2.3— Continued 

Location in SL 
or other key 

Main questions 
in proposal 

Subthemes/research questions to explore 

What is the significance of landholding or use arrangements (size, quality, ownership, other)? 
What vulnerability issues may keep people from adopting or stop them from using technology 

(natural; health crises [AIDS, malaria, and other]; retrenchments; changing family relationships/ 

composition; market changes/prices)? 
How is soil fertility perceived? How identified and assessed? How is fertilizer perceived? Is 

it money or affect on soil, and so on? How does it affect qualities of crops (role of myth; 

magic; science, and so on)? 
How does SFR mitigate vulnerability? Are poor people more or less vulnerable to vulnerability 

factors (pests, late planting, striga, food shortages)? Compare SFR to other kinds of choices 

that farmers could make. 
What assets do people have access to that are not being used and how could they be used? 

Structures and 

What is the role of policy? Are farmers even aware of the policies (local)? Are the disseminators 
aware? How do these policies facilitate or constrain them? What is the enabling environment 
— what do farmers want them to do? Are there competing policies (land use; soil fertility; 
conservation; cost-sharing: health, schools)? Who is enforcing them and whom is it being 
enforced upon? 

Assets and 

4. What are the effects 
of the SFR inter- 
ventions on people's 
productive and risk- 
mitigating assets? 

How do these technologies facilitate people's investment in assets (physical, natural, social 
capital, human, financial, cultural)? 

Does it increase labor availability? 

What is the relationship between adoption of SFR and crop productivity (differentially defined)? 
In relation to different crops and relative importance of those crops? What maize varieties 
do better with SFR technologies? What other qualities does SFR help: taste, color, drought 
resistance, milling, food security? 

Are the practices sustainable (physical and financial)? How do people perceive the sustainability? 

Does it strengthen or undermine community and household cooperation (for example, between 
poor and rich)? Does use or success of technology increase or decrease sharing or individuality 
(sharing of knowledge, wood, seeds; fighting striga; protection from mud damage; scaring 
of birds; scrambling for tithonia; refusing to share seeds, etc; husbands and wives working 

Does the technology help people to diversify activities and cope with weather, threats to 
economic shocks, or trends? Or does it increase vulnerability? 


Table 2.4 Research design matrix: Dissemination strategies 

Location in SL 
or other key 

Main questions 
in proposal 

Subthemes/research questions to explore 

structures and 

5. What are the main 
dissemination approaches 
being used by different 
government and non- 
government institutions 
and what levels and 
types of participation by 
farmers do they involve? 

What are the main categories of dissemination approaches? 

What are the relative financial costs of these to the institutions? 

How do people value the time spent at meetings, farmer field visits, and so on. What are the 

economic and social costs and benefits? Do they make contributions? Why do they go? 
What do you get out of it? Opportunity costs of not going? 
What are the informal mechanisms through which people learn about these technologies? 

Observation, social relationships/social capital; learning by doing 
What were they doing before? What kind of experimenting? What kinds of soil fertility 

management? Are they still using them? Stopped using them? Combining them? 


6. How effective are 
these approaches at 
reaching the poor? 

Under these approaches, to what extent are different categories of farmers (for example, very 

poor, poor) being reached with information? 
What kinds of information do they expect to get? Why do different types of farmers come? 
What groups exist previously related to agriculture (map picture to compare)? 

TSP livelihood 


7. How do these 
approaches, including 
their principles, organiza- 
tional forms, method- 
ological tools and 
processes, impact human 
and social capital 

What have been the effects on social capital formation or strengthening and why? How do 

these differ among different social groups (among the poor, ethnic groups) 
Under these different approaches, what kinds of innovations are emerging? 
What have been the effects on human capital formation and why? (Measure training given 

and knowledge acquired.) 
Have people taken on new activities through their organizations/groups? 
What institutions and services do people know about? 
Have they increased their demands on outside institutions (nongovernment organization, 

Ministry of Agriculture at local level, ICRAF/KEFRI/KARI, banks, AFC) or ability to 

negotiate with them? Which institutions do they use? Also ask about approaching political 

To what extent to they make demands or seek information/assistance from village 

committees, the sub-location committees; local authorities (councilors) or other 

government officials; the chiefs 
Which institutions do they find most helpful and why? What do people want to get from them? 
Do they alter or reinforce power relationships in communities? 
How effective are these methods at creating a sense of ownership and commitment to the 

Have the groups done any further dissemination in other villages? What were the 

experiences of this group in doing this dissemination? 
How have practices of extension agents (including ICRAF/KEFRI, government, and other) 

affected local social relationships (are farmers who are visited frequently seen as "model 

farmers," respected or resented? Does the attention make people want to work harder and 



8. What key factors 
explain why these dif- 
ferent approaches are 
more or less effective in 
making an impact on the 

What is the relative importance of social relationships (including gender, age)? Why are 

certain approaches more or less effective for women and men? 
Levels of poverty /wealth: Why are certain approaches more or less effective for poor and 

non-poor farmers; other groups (method for choosing participants; expectations of 

What is the role of cultural and ethnic variations (including trust, cooperation, preferences 

for individual or collective activities)? 
Human capital (education) 
Factors in the institutional environment (including relationships and coordination between 

different institutions) 

(continued ) 


Table 2.4— Continued 

Location in SL 
or other key 

Main questions 
in proposal 

Subthemes/research questions to explore 

What are people's perceptions of the different institutions and their approaches, and the 

individuals involved? 
Farmers' perceptions of the individuals responsible for dissemination? 
Quality of training (too little or too much information; appropriateness; equal partners or 

students/do ideas flow both ways; follow-up) 
To what extent have policies, practices, and training been changed in response to information 

from farmers? 
Methods and tools used: language, teaching materials 
What is the significance of the early stage of the approach? (Was method developed with 

people or presented to them? How much work was done in advance to understand local 



The Context of the Research 

The aim of this chapter is to contextualize the relationship between technology and de- 
velopment processes with respect to this research, dealing with social and cultural re- 
lationships, customs, and institutional dimensions of social and economic life in the 
region. Processes such as socioeconomic differentiation and gender are crucial to understand- 
ing the importance of power, how power works, and how it shapes decision-making. The 
chapter also draws attention to the agroecological conditions (climate, soils, rainfall patterns) 
for farming; the skills of local people to generate a description of human capital; issues of 
poverty; and the agricultural policies implemented by the Kenyan government. 

It is crucial to note that the research area is inhabited by two different ethnic groups: the Luo 
and the Luhya. Where relevant for the research, the differences between them are elaborated. 

People, Institutions, and Structures 

This description is based on a literature review and some qualitative fieldwork. While distinct 
differences remain between the Luo and Luhya, over the years the cultural repertoires have 
changed due to interactions with other cultures, modern and colonial laws, and customs 
of social order. 

Between the Luo and Luhya, social life hinges around kinship relationships that define 
rules and obligations vis-a-vis each of the individuals belonging to a certain social unit. This 
social unit is at best represented at different levels: the "household," the family, the clan, and 
the lineage. These are reflected in settlement patterns as well, such as the compound and the 
village. The compound is something typical for the Luo and is less important and visible in 
the Luhya context. Furthermore, it appears that in the Luhya context of decision-making, the 
"household" plays a larger role than among the Luo. Here the extended family plays a more 
important role, particularly in situations where the son(s) are still residing in their father's 
compound. As will be explained later, genealogical seniority is very important among the Luo. 

A typical Luo homestead (dala) consists of a site with the houses of monogamous or 
polygamous domestic groups, surrounded by their fields. The smallest social unit in the 
homestead is the "household," which is made up of at least two generations: the father and 
mother(s), and their offspring. (The "household" as considered in the qualitative research may 
also include others with significant livelihood linkages, for example, nonresident sons, and 
sometimes non-blood relations.) Occasionally, households of brothers of the homestead's 
owner are also found there, as well as servants and "strangers." Several homesteads make up 
a village. Residence is based on kinship, but also on alliances developed out of strategic con- 
siderations. The elementary social relationships among the Luo and Luhya are organized 
around the normative principle of patrifocality that cements the relationships between father, 
mother, and their children. This unit thus deals with people of the same father and operates as 



one corporate group and shares and dis- 
tributes most of the domestic activities. 
Daughters are included only before mar- 
riage, since they are not considered when 
it comes to the inheritance of wealth. The 
normative respect for age (that is, seniority) 
is such that the eldest son must marry first, 
then the second eldest, and so on in order of 
seniority; the same is true of daughters. 

When the senior son of a Luo father 
marries and has children, he is the first to 
build a new and independent homestead. 5 
When the father dies, the eldest son takes 
over the responsibilities of leadership of the 
family. In situations of polygamy, relation- 
ships then start from the matrifocal unit that 
combines a mother, her sons, and unmarried 
daughters as an independent set of people. 
This implies affiliation to the mother rather 
than to the father per se. In a monogamous 
situation, the position of the father is very 
strong, as there is no rivalry. In a polyga- 
mous situation, the position of the father 
is weakened substantially in favor of the 
mothers and grandmothers. The relationship 
between the matrifocal units within the 
compound is referred to in terms of "jeal- 
ousy" when it refers to the relationship 
between co-wives themselves and "rivalry" 
when it involves all in a matrifocal unit as a 
group against another, opposing group. 6 The 
matrifocal unit that combines a mother and 
her sons in the second generation is referred 
to as "the people of the same grandmother." 
At this level, the rivalry and competitive re- 
lationships between co-wives and their sons 
start fading. The position of the grandfather 
regains importance. Beyond the grand- 
mother and grandfather line, at the third 
and up to the fifth generation, the extended 
family appears as the next organizational 
form. People descending from the same 

great-grandfather make up extended and 
often polygamous families. The elders act as 
representatives in disputes between oppos- 
ing families. They are also intermediaries 
between younger members and the an- 
cestors and therefore act as foster-father 
guardians. They form the first organized 
council to arbitrate land and boundary dis- 
putes between members of their extended 
family. At this stage, social control of the 
community is exercised partly through the 
authority of these elders and partly through 
the control over the means of accumulation, 
which the leader of the group protects. Con- 
trol and accumulation of resources is a basic 
requirement for subsistence and competi- 
tion in Luo and Luhya society. 

A next level is the lineage, which in- 
volves descendants of a common ancestor, 
usually from four to seven generations back. 
It is a maximal lineage of landholding, co- 
operating agnates and generally considered 
to be the backbone for settlement, house- 
hold and family formation, and social re- 

When it comes to marriage, both kinship 
relations and the seniority principle are of 
primary importance. Marriage is arranged 
between families of different clans. The Luo 
custom is that the senior son of the senior 
wife should marry first. When he is of age, 
he is first given a cow and a young bull, 
which lineage members take to the bride's 
homestead. Bride wealth is only drawn 
from the male lineage, particularly from the 
grandfather's or great-grandfather's line. 
Daughters of the same mother also marry in 
order of seniority. The seniority principle is 
less prominent among the Luhya. 

Land is not private property per se. De- 
spite title deeds issued over the years to the 
customary male landholder, it is hard to 

5 The Luo call this "liberation." They distinguish, however, between two different forms. The first liberation 
is when a woman starts cooking in her own house in the compound of her father-in-law. The second liberation is 
when a man establishes his own homestead. 

6 This relationship often generates the various kinds of conflicts, competitions, envy, confrontations, and divisions 
that are so characteristic at various levels of Luo social organization. 


speak about private land and/or a land mar- 
ket. Land distribution is still very much 
shaped by the image that it is the family and 
the lineage or clan that controls the land. 
Land inheritance is still basically organized 
around father-son(s) relationships. The 
mondo, the father's field, is inherited by the 
last-born son. Women are not considered 
in land inheritance or allowed to own 
land. Women cultivate the land, however. 
Monogamy and polygamy make a differ- 
ence here, although there is variation from 
the general rule. Among the Luhya, when a 
father dies, land is subdivided between the 
mother and the sons; each has his and her 
own plot and own gate. This differs quite 
substantially from the Luo. When the hus- 
band dies, the wife cultivates smaller parts 
of the son's land. Another major difference 
between the Luo and Luhya is that, among 
the latter, there is much more evidence of 
land transactions, including purchasing and 
renting. A further difference is that the 
Luhya practice circumcision, unlike the Luo. 
The lineage and clan also play an im- 
portant role in the political life of the Luo 
and Luhya. The clan-head is someone who, 
together with the elders, rules with refer- 
ence to his understanding of customs and 
customary law. The clan-head, however, 
does not denote a clearly defined office en- 
tailing specific rights and duties. The status 
of a clan-head is based more on personality, 
influence, and personal status; he is some- 
one who stood out among his age mates. 
Currently, the status of clan-head has dimin- 
ished in favor of village elders (who are not 
necessarily the oldest in the village), chiefs, 
assistant chiefs, and elected councilors. Be- 
longing to the country's political parties, 
councilors have roles, duties, and responsi- 
bilities fixed by public law. Chiefs and as- 
sistant chiefs are appointed by government 
and derive a salary from that office, unlike 
village elders, whose influence depends on 
status within the village. The chief orga- 
nizes his public meetings (barazas) to dis- 
cuss particular issues with the people and 
often provides an entry point for NGOs and 

government institutions into the commu- 
nities. Clearly the de facto authority of the 
chief varies from community to community, 
as some barazas are poorly attended. 

Social Categories: 
Gender, Age, and Class 

Based on the preceding description, the most 
important and relevant social categories to 
consider are gender and age. Roles, tasks, 
and responsibilities of men and women are 
fairly clear and respected. Women do not 
own or control land but provide most of 
the agricultural labor — planting, weeding, 
harvesting — and perform most domestic 
tasks — cooking, fetching water and fire- 
wood, washing, raising children, and so on. 
Men do the heavy work (plowing) and de- 
cide how to use the plot and spend the 
money earned. But we have to be careful 
with such a conclusion. First, this does not 
hold in many cases where men are absent, 
as when they work outside the village. Sec- 
ond, if we listen carefully to women them- 
selves, they indicate their ability to decide 
for themselves and to negotiate with their 
husbands to secure most of their needs for 
food, money, and protection. Not in all cases, 
however, do women successfully negotiate 
with their husbands. 

The case studies point up that women 
make many decisions, as they do most of 
the farmwork; however, few appear to 
notice or acknowledge this role. Typically, 
they consult with their husbands after actu- 
ally making the decisions. For example, in 
the case of soil fertility replenishment tech- 
nologies, it was evident that women fre- 
quently make the decision to implement or 
not. This is possible because they are the 
laborers on their own farms. For instance, 
Mary, one of the farmers, was the first in the 
household to receive knowledge about use 
of tithonia and the fallow trees. She dis- 
cussed this with her husband when she had 
planted and tried them out on her plot. 
Agnes works alone because her husband has 
no interest in working the farm. Therefore 


she decides on what technology she wants 
to employ. Clearer still is the case of two 
co-wives from the same homestead and 
husband. Florence decided to adopt the SFR 
technologies, while Patricia did not, each 
for her own reasons. 

An emerging social category is that re- 
ferred to as the female- (or even child-) 
headed household. Either because of a 
husband's long absences, divorce, or death, 
households are increasingly managed by 

Age also plays a major role. Generally, 
the elders control village life politically and 
take responsibility for maintaining tribal 
norms and values (such as respect for 
elders) and customary laws. An implica- 
tion of the responsibility and prestige of 
genealogical seniority among the Luo is 
that it puts elders in the position of first har- 
vesting (dwoko cham), and first sowing (golo 
kodhi), as well as of eating specified parts of 
animals, usually the best meats. In these sit- 
uations, it is the elder who determines what 
and when to plant. Similarly, in a polyga- 
mous setting, it is the first wife who must 
start plowing and planting first before the 
second wife. If these rules are not obeyed, 
chira will be her fate, as one informant ex- 
pressed. Chira is a curse that befalls a per- 
son who has violated some taboo. 

A third level relevant for discussing de- 
velopment and change is one that deals with 
processes of social differentiation. Land, 
being an important asset, is not equally dis- 
tributed. Only some of the individuals in our 
study "own" more than 10 acres of land; the 
majority own zero to five acres. This quan- 
titative aspect of landownership, size, is 
important in the analysis of technology and 
social change. 

A second element that plays a role in so- 
cial differentiation is related to income gen- 
eration, even in a largely subsistence econ- 
omy. The amount of land and the ability to 
control and mobilize family and/or hired 
labor plays a role in the amount of money 
people earn in agriculture. Income in the 
form of remittances from labor migration 

also plays a major role. Access to land, 
labor, and urban jobs is clearly associated 
with income differentiation. 

Agroecological Conditions 

The research in western Kenya is focused 
largely on medium-to-high potential high- 
land areas. Rainfall is good, ranging from 
1,200 to 1,800 millimeters per year with two 
cropping seasons annually: the long rains 
from March to July, and the short rains from 
August to November. The short rainy sea- 
son is traditionally less reliable in terms of 
total rainfall and length of growing season, 
but the rains have been good since 1998. 
Rains are slightly less in Busia, one of our 
sites nearer the Ugandan border, and no- 
tably less in another site, Rachuonyo District 
to the south. The altitude ranges between 
1,200 and 1,700 meters above sea level and 
the topography is undulating with moderate 
slopes. Soils are of generally good physical 
structure but are low in nutrient stocks. In 
many parts of the region, phosphorus is the 
major limiting nutrient, but nitrogen and 
potassium limitations are also prevalent 
(Shepherd et al. 1996; Jama et al. 1998b). 
Soils are much less fertile in the lower el- 
evations (such as Rachuonyo). Moreover, 
heavy infestation with Striga hermontica, a 
parasitic weed that devastates the maize 
crops, is common (Oswald et al. 1996). 

Socioeconomic Context: 
Vulnerability, Education, 
and Poverty 

High population densities prevail, ranging 
from 500 to 1,200 per square kilometer in 
all of the western Kenya sites, and are par- 
ticularly acute in Vihiga District. The Luhya 
inhabit Kakamega, Vihiga, and Busia dis- 
tricts, while the Luo reside in Siaya and 
Rachuonyo districts. The farming system in- 
corporates crops, livestock, and trees. Maize 
(local varieties) and beans are the most 
common agricultural products. The food sit- 
uation was reported as deficient by 89.5 per- 
cent of the households in Siaya and Vihiga; 


they had to buy food to supplement their 
own harvest (Wangila, Rommelse, and De 
Wolf 1999). Only 8.9 percent of the house- 
holds were food secure from their own pro- 
duction. Average household income for west- 
ern highland households was only US$1,014 
and crop income was a paltry US$321, ac- 
cording to a recent study (Argwings-Kodhek 
et al. 1999). Average labor productivity from 
agriculture was about US$76 per year in 
western Kenya, only one fourth the level 
achieved by farmers in central Kenya. 

In fact, many of the communities under 
study are among the poorest in all of Kenya 
and clearly the poorest among the medium 
to high potential areas. For example, a re- 
cent national study of poverty found Western 
Province (including Kakamega and Vihiga 
among its four districts) to be one of the 
poorest in the country (Kenya — Ministry of 
Planning 2000). It was estimated that 31.5 
percent of households in western Kenya 
are among the hardcore poor, as opposed 
to 19.6 percent for all rural areas. Western 
Province and Nyanza Province (including 
Siaya District) also had high incidences of 
sickness that were twice as high as those 
reported in central Kenya, an area with 
similar farm sizes (median of about one 

Agricultural Policies, 
the State, and NGOs 

Agricultural polices of the state for the 
country and region under study rests on 
providing services and incentives to agri- 
cultural producers in order to increase pro- 
duction to satisfy a variety of needs: that of 
the economy as a whole (export and na- 
tional consumption) and for individual 
households. Policy objectives were aimed 
at stimulating exports from industrial crops 
in certain regions and to promote food pro- 
duction (mainly maize) in the much larger 
remaining areas. The government hoped to 
achieve this through a mixture of market- 
and non-market-oriented instruments. From 
the colonial period onwards, state efforts 

were geared toward setting up relevant in- 
stitutions (extension, technology research 
and development, markets for agricultural 
commodities, physical inputs, capital, and 
land). These state structures are many — 
there were 57 parastatals on the books of the 
Ministry of Agriculture and Rural Develop- 
ment in 2000. But over the years, their pres- 
ence and influence have diminished. 

Before Structural Adjustment, the state 
was almost omnipresent as an extension 
agency supplying and disseminating knowl- 
edge and new technologies, as a credit 
agency supplying credit and loans to farm- 
ers, and as a market agency determining 
input and output prices (of seed, maize, fer- 
tilizer, milk, and so forth). It also controlled 
marketing as the sole buyer of agricultural 
commodities. This particular role of the 
state was not without problems. It is widely 
known and referred to in both the literature 
and the field that these state services were 
very inefficient. State interventions that 
attempted to improve peoples' livelihoods 
were plagued by corruption and failing 
markets. This is especially true of the output 
side, where the prices paid to farmers were 
notably low and delayed. 

Structural Adjustment, implemented as 
of the early 1990s, was launched to stream- 
line the functioning of the agricultural sector. 
Structural Adjustment spurred political and 
economic reforms such as privatization and 
liberalization. Many government institutions 
vanished; others now face competition from 
private companies. In the process, extension 
services to farmers have substantially dwin- 
dled, and agricultural credit supply lies ex- 
clusively in the hands of private institutions, 
although some NGOs operate microcredit 
schemes in the region. The previously exist- 
ing tractor-hire scheme and fertilizer sub- 
sidies were already abolished long before 
structural adjustment programs were im- 
posed on the country. 

Now that many markets are liberalized 
and decontrolled, agricultural markets have 
not become less efficient but have created 
unequal access to inputs and revenue. The 


private sector has responded well to provide 
trade opportunities in the more productive 
regions of Central Kenya and the Rift Valley 
(e.g., in the dairy sector), but have proved 
inadequate in the western highlands. Many 
high civil servants were appointed as man- 
agers of newly privatized businesses; be- 
cause of corruption, various commodity 
markets and cooperatives have started to 
collapse, with private monopolies emerging 
in these less favored areas. For the western 
Kenya region, the operation of input and 
output markets was seriously affected. The 
once booming sugar industry, coffee sector, 
and textile industry have collapsed almost 
completely as a result of these processes. 
Regulation of the private sector is poor as 

well, and farmers register frequent com- 
plaints of fraudulent activities, such as adul- 
teration, by retailers of seed and fertilizer. 

The advent of structural adjustment 
made NGOs more prominent in the field of 
agriculture, however. Numerous NGOs now 
render services to farmers. CARE-Kenya 
(working with smallholder farmers on adap- 
tive research for maize and soil fertility, 
health, and education), Lagrotech (provid- 
ing open pollinated maize varieties), and 
OFPEP (an environmental NGO concerned 
with issues of soil fertility and erosion) are 
all examples of NGOs becoming more 
prominent now that the state is less and less 


Poverty in Context 

The aim of this chapter is to discuss poverty in western Kenya in detail. The first section 
deals with official government definitions of poverty. The second elaborates how 
ICRAF conceptualized and understood poverty in its technology-development work, 
monitoring and evaluation exercises, and quantitative analysis. The third section draws on the 
qualitative research and attempts to depict how the various people in the communities define 
poverty. The last section, based on qualitative and quantitative data, explores who the key or 
most vulnerable groups are among the poor. 

Official Definitions of Poverty 

The Government of Kenya undertook a nationwide study of poverty (Wealth and Monitoring 
Survey) in 1994 and 1997. It is generally agreed that the 1997 data collection exercise was 
more accurate than the one in 1994, so it is these data that are used to estimate poverty levels 
across the country. To calculate poverty rates, the government uses a headcount of persons 
below a specified poverty level and then divides this by the total population. The poor are de- 
fined as those members of society who are unable to afford minimum basic human needs, com- 
prised of food and non-food items. Food requirements are based on World Health Organization/ 
Food and Agriculture Organization of the United Nations (WHO/FAO) minimum standards 
for food energy intake (FEI) of 2,250 calories per adult per day. This is then translated into an 
expenditure value of 927 Kenya shillings (KSh) per adult per month for the rural areas. 7 Non- 
food requirements are estimated at 312 KSh per adult per month. The absolute poor include 
anyone under the level of 1,239 KSh per month and the hardcore poor include anyone under 
927 KSh per month. 

In 1997, it was estimated that the percentages of the population in hardcore poverty were 
the following for western Kenya districts: 

Percentage in 


hardcore poverty 













7 There are slight adjustments by province to account for purchasing power disparities across rural locations. 



While some reliable quantitative panel data 
are available from other African countries, 
this is not yet possible in Kenya using offi- 
cial statistics. Some other studies are useful 
in assessing changes in poverty indicators 
(e.g., Jayne et al. 2001). 

ICRAF and Conceptions 
of Poverty 

ICRAF conducted a number of diagnostic 
studies around field sites in east and south- 
ern Africa during the late 1980s. These 
aimed to identify farmers' main problems 
and then design agroforestry systems that 
could address them. Not surprisingly, low 
resources and incomes emerged as key so- 
cioeconomic problems in all the sites. At the 
same time, different biophysical problems 
were found across the locations: poor soil 
fertility, excessive soil erosion, lack of wood, 
low quality fodder, and unmet demand for 
fruits. Initial technology design processes 
were focused more on the technical aspects 
of meeting these biophysical needs rather 
than in addressing the socioeconomic needs 
and constraints. Thus a phase of researcher- 
managed trials ensued with relatively low 
input from social scientists. 

This all changed in the mid-1990s, when 
farmer-designed and managed research be- 
came the most important mode of exper- 
imentation. In addition to making strides 
in advancing technological designs, this 
brought the socioeconomic constraints and 
needs of farmers to the forefront. Thus, the 
major conceptual lens for western Kenya 
evolved from a total focus on low soil fer- 
tility to one that put more and more em- 
phasis on poverty alleviation. Despite this 
evolution, the focus on soil fertility did not 
change, as it was felt that poverty could 
not be reduced without increases in soil 
fertility and thus agricultural productivity. 
But further probing into farmers' resources 
reveals that poverty implied certain con- 
straints. For example, farmers would not be 
able to purchase seedlings, so soil fertility 
technologies that relied on seedlings would 

not be appropriate. Small farm sizes also 
meant that reliance on a single technology, 
such as improved fallow, would have seri- 
ous limitations. 

In terms of ICRAF's research process, it 
embraced the goal of poverty reduction and 
adjusted its research strategy accordingly. 
The first step was to understand who the 
"poor" were. Wealth-ranking exercises were 
held in many pilot project villages to iden- 
tify the criteria used by villagers to classify 
themselves as better off or worse off. The 
villagers also placed themselves in their 
own categories (between three and five, de- 
pending on the village). ICRAF then used 
some of these criteria to make its own quick 
assessments of where households stood 
along the wealth-poverty continuum. These 
criteria included farm size, number of live- 
stock held, whether fertilizer is purchased, 
and whether labor is hired. ICRAF scien- 
tists used this information to identify targets 
for impact assessment and to stratify house- 
holds for testing and monitoring work so 
that the poor would be involved in technol- 
ogy development. ICRAF also studied the 
links between perceptions of wealth and so- 
cial networks to understand better how the 
poor could be reached. 

Local Definitions and 
Understandings of Poverty 

This section draws on the qualitative data 
gathered through the case studies and the 
group discussions. For these discussions, 
men and women sat separately with the 
fieldworker. Poverty appears to be a slip- 
pery concept. Most commonly heard was 
the perception that "nobody is poor." The 
notion of "poor" or poverty, rather, repre- 
sents a rejected type of person. It was thus 
reported that "Poor people are those that 
are handicapped. The poor stay and beg in 
towns, as they do not have land and shelter. 
At least I have a shelter." 

There is no "rich" person either. The 
"rich" were described as those who have 
something extra. Since nobody in our sample 


was seen to actually have anything extra, 
people reason that there are no rich people. 
The notion of "rich" is not favored or used 
at all in everyday discussions. According to 
general standards, the rich would not admit 
they are rich. Nevertheless, it is widely ac- 
knowledged that the "rich" have more land 
than the "poor." They therefore have more 
to leave fallow and thus could profit more 
from improved fallow technologies. 

Reluctant to classify themselves or others 
as poor, people prefer to say that they are 
"lacking something." Poverty is thus asso- 
ciated with lacking income both from em- 
ployment and from business. 

While classification along poverty lines 
is contested, differentiation among mem- 
bers of the community is acknowledged. 
People will still describe characteristics of 
being poor. Poor people are those who have 
small pieces of land, grass thatched houses, 
and large families with children walking 
in tattered clothes who have fallen out of 
schools. Poor people engage in casual jobs 
and can never buy enough food. Physical 
disability is seen as a cause of poverty; 
many of the physically challenged beg at 
the market. Mental disorders are very com- 
mon among the poor because they are un- 
able to seek assistance for the condition. 
Poverty was said to refer to: 

• Lack of land 

• No daughter or son on the farm 

• Inability to feed one's family 

• Inability to pay for education, health 
care, and so forth 

• Wearing tattered clothes 

• Having unemployed children 

• Physical disability 

• Housing with a leaky roof 

Laziness and drunkenness are often 
cited as causes of poverty within the com- 
munity. Nevertheless, there is more evi- 
dence that unemployment and increase in 
population are the real major causes of 
poverty, since there is no industry in the 
area, and little diversity in business to gen- 
erate income. 

Rather than speaking of the "rich," peo- 
ple spoke of an ideal type, a "good farmer." 
A good farmer, respondents said, seems to 
be able to combine the good things of life. 
A good farmer has high crop yields, his 
stores are always full of maize, and he has 
zero grazing animals. The farmers claim 
that a good farmer never buys food. Instead, 
people flock to his farm. A good farmer 
uses hybrid maize varieties such as 511, 
farmyard manure, diammonium phosphate 
(DAP) fertilizer and top dressing using cal- 
cium ammonium nitrate (CAN), and urea. 
Therefore, a good farmer is one with the 
money to buy these inputs. 

We have to take into account that in com- 
munities such as these, most women and 
young adults are the active farmers. From 
observation, women weed, prepare farmyard 
manure, and milk cows. Women are also 
more likely than men to take casual jobs. 
According to the women, most men are far 
from home and send too little money to 
sustain the family for a month. Those men 
who are present are either "too old" or have 
business in Luanda, a market town in Siaya 
District, and are hardly seen on the farm. 
Women claim that their husbands argue that 
their work is to pay school fees and under- 
take major projects such as constructing 
houses and purchasing land. 

Milking cows and growing vegetables, 
such as kale, cowpeas, and groundnuts are 
considered to be petty activities that belong 
to women. Many newly married women 
have no options, as many of their husbands 
engage in formal employment or small- 
scale business outside the home. It is still 
customary that women should stay at home 
and maintain the home. Even if a woman is 
working or has a business, she should return 
home early. 

A second area for analysis is what 
makes people vulnerable. The following are 
named by local people as contributing to 
vulnerability among the poor: 

• Limited principal resources/assets such 
as land 

• Unemployed children 


• Theft 

• Lack of alternative to farming as a 
source of income 

• Lack of markets 

• Low prices 

• Being childless 

• Old age 

• Illiteracy 

• Poor health, especially HIV/ AIDS 

The vulnerabilities listed are intrinsically 
the outcome of social processes. Through 
an analysis of such processes we can better 
understand the effects of agroforestry tech- 
nology on social development, linking the 
social, political, economic, and agroecolog- 
ical context in which all this occurs. These 
social processes can be broken down as 

• Some of the vulnerabilities refer to a 
lack of an enabling institutional envi- 
ronment. The way the market operates, 
and specifically prices for commodities 
in relation to the cost of inputs, makes 
commercial agriculture uneconomical. 
The relevance for soil fertility is that 
the fertilizer is beyond almost any- 
one's financial reach. Such a situation 
strengthens the need for alternative soil 
fertility enhancing technologies. 

• Other vulnerabilities, on the other hand, 
refer to other externalities, such as 
AIDS, that both reduce labor and 
drain financial, physical, human, and 
social assets. In conditions where labor 
and financial resources are drained, in- 
troducing labor-intensive technologies 
may be counterproductive. 

• The vulnerabilities listed also point to 
the overall importance of accessing and 
mobilizing resources (social, financial, 
natural, human, and physical) and the 
character of such resources. 

• More specifically, some of the vulner- 
abilities hint at issues of aging. Intro- 
ducing labor-demanding agroforestry 
technologies to an aging community 
puts pressure on the availability of 
labor at community and family levels. 

The implications of aging need to be 
understood at the level of livelihoods: 
many people would rather migrate to 
urban centers in search of jobs rather 
than retain a land-based livelihood with 
low returns to labor given low market 
• The vulnerabilities listed also draw at- 
tention to issues of security (e.g., theft) 
and incidences of diseases that affect 
both plants (Striga) and humans 

Findings, however, also show that inci- 
dences and severities of poverty are cyclic, 
especially between generations, and this pat- 
tern, too, tended to follow in the area of 
adoption and dis-adoption of the technolo- 
gies under consideration. Whereas there is 
sufficient evidence that SFR technologies 
adopted have impacted positively on the 
lives of the poor, it is also clear that some 
of the basic requirements of these technolo- 
gies have continued to exclude some of the 
most vulnerable members of the commu- 
nity. From the case studies, the biggest risk 
is hailstones and heavy winds that crush 
down maize. Many case study farmers cite 
the lack of money as another vulnerability. 
Because of lack of money, they cannot hire 
labor for tithonia use or to dig terraces, can- 
not buy pesticides and insecticides for 
tomato growing (this is one of the reasons 
why tomato growing is not very popular — 
they claim tomato blight can clear the whole 
crop, which leads to losses). They also lack 
agricultural assets, for example, ox plow 
and tractors for faster tilling. This puts addi- 
tional burdens on labor and requires cash 
payments (about 50 KSh per person per 
day) and to supply meals to the laborers 
when they work a full day (about four to six 

Striga is also very common and causes 
stunting in crops. Streak virus is a disease 
they are not familiar with and turns leaves 
yellow, then they rot and dry. Cassava mo- 
saic has cleared cassava in the village; one 
cannot find a single cassava stem in the 


village. Moles have worsened the cassava 

Lack of food is the main worry for most 
of the community members, as without food 
— defined as maize — they cannot have the 
energy to work. Lack of food is equally a 
problem for the casual laborers, as work is 
intermittent and pays only about 50 KSh a 
day. During the months of April, May, and 
June, maize prices shoot up to between 15 
and 20 KSh per kilogram. By observation, 
many people seem to work or farm just to 
get something to eat. 

Vulnerability is compounded by disease. 
During the heavy rains, malarial mosquitoes 
increase, and hungry people have lower re- 
sistance to malaria. The incidence of death 
is high during this period, drawing more 
time away from farming. From the case 
studies, HIV/ AIDS is also a major threat to 
the society. It robs families of strong young 
men and women, leaving behind elders who 
cannot handle heavy work and orphans who 
need to be fed and educated. On other as- 
pects of livelihoods, the fluctuation in prices 
of farm produce is a further concern. During 
the harvest, maize prices drop as low as 10 
KSh per a two-kilogram tin, while during 
planting time, prices shoot up to 30 KSh 
for the same amount. In addition, Kenyans 
from Kitale, Molo, and Naivasha flood the 
market with cheap maize, tomatoes, cab- 
bage, and kales (sukuma wiki), making the 
crops that local farmers have toiled for lack 
market value. 

Key Groups among 
the Poor 

In identifying the poor, we have not used a 
simple definition, but rather have attempted 
to employ alternative approaches in attempt- 
ing to distinguish the poor from the less 
poor. As noted earlier in the methodology 
section, the first method used by ICRAF 
was that of wealth ranking. During this 
exercise, villagers identified criteria that 
they found important in differentiating be- 
tween socioeconomic classes. A few of these 

variables — those that were more robust 
across sites and easily measured (farm size, 
cattle holdings, use of fertilizer, hiring of 
labor) — were used to create a wealth index 
for each household in the pilot villages. This 
wealth variable was then used as the poverty 
marker for households in the adoption study 
within the pilot areas (n = 1,633). The draw- 
back of this variable is that there is no stan- 
dard of comparison with which to judge 
whether or not a particular household is 
poor — it only provides an ordinal ranking of 
households. Therefore, it is not possible to 
propose that a certain percentage of house- 
holds are "poor." For a subsample of pilot 
village households (n = 120), more detailed 
information is available. We can further 
classify households on the basis of asset 
poverty, expenditure poverty, or food con- 
sumption poverty (measured at the begin- 
ning of the study period). Each of these 
methods is used. For assets, we then calcu- 
lated the number of months a household 
could cover its basic subsistence needs if 
all non-land, non- building assets were liqui- 
dated. We assumed that coverage for less 
than three months constituted extreme 
poverty. The middle group could sustain it- 
self for between 3 and 12 months, while the 
well-to-do group could meet its needs for at 
least one year. For expenditures, data were 
collected on spending over a three-month 
period. Because that period is not neces- 
sarily representative of spending at other 
times of the year, we do not attempt to gen- 
erate an annual expenditure amount. 

Thus, we have calculated per capita ex- 
penditures and distinguished three groups of 
households based on inductive discovery of 
gaps in the distribution (this was at 800 and 
2,500 shillings). For food consumption, we 
are able to make some normative- based 
calculations of the number of people in 
poverty, based on minimum requirements 
for food and basic needs (e.g., see the Gov- 
ernment of Kenya definition in this chapter). 
Households not meeting minimum daily re- 
quirements are treated as poor, while those 
who meet minimum needs but do not exceed 


Table 4.1 Distribution of poverty in pilot villages (// = 104) using 
alternative classifications 

Percentage in wealthiest group 
Percentage in middle group 
Percentage in poorest group 

Months sustained 
by assets 


per capita 








150 percent of the minimum are considered 
to be in the middle group. 

In all three cases, there is a sizeable per- 
centage of households in the poorest class — 
as many as two thirds using the asset mea- 
sure (Table 4.1). While this is reasonably 
consistent across measures, the subjectivity 
of the measures reveals itself more clearly 
when attempting to identify a wealthier 
group. Estimates range from as few as 2.9 
percent to as many as 26.9 percent of house- 
holds. Of course, such variation in estimates 
may be valid, since we cannot expect all 
poverty indicators to be highly correlated. 
Analysis using all three measures shows 
that 19.1 percent of households are consis- 
tently classified as poor. A further 38.4 per- 
cent fall into the poorer group in two of the 
three measures so a total sum of 57.5 per- 
cent are estimated to be poor by at least two 
of the measures. At the other end, no house- 
hold is classified as wealthy under all three 
measures and only 5.7 percent are estimated 
to be wealthy in at least two measures. 

In the non-pilot villages, our cross- 
sectional survey attempted several ways of 
distinguishing the poor from the non-poor. 
In terms of quantitative measurement, a 

detailed asset inventory was obtained and 
related to household size and basic needs, 
similar to that for the pilot villages. We also 
tried to measure poverty from two addi- 
tional perspectives. The first was based on 
household self-perception in which respon- 
dents were asked to compare themselves 
to their neighbors on a number of wealth- 
related criteria (e.g., yields, off-farm income, 
ability to cope with risks, natural capital). 
From this, an index was created. Based on 
the distribution of the index, we were able 
to isolate a group of households that were 
better-off than the rest. The cutoff between 
the lowest and middle groups was more 
arbitrary, but was selected so that those in 
the lowest group in fact ranked themselves 
as worse off than their neighbors in most 
categories. The second identifier of the poor 
was an assessment by enumerator after hav- 
ing inspected the entire farm. The outcome 
was the classification of the household as 
either very poor, poor, or non-poor. The 
results of these classifications are presented 
in Table 4.2. Between 9.4 and 17.5 percent of 
households are identified as being wealthy 
relative to their counterparts. The lowest 
estimation of the number of non-poor came 

Table 4.2 Distribution of poverty in non-pilot villages (n = 360) using 
alternative classifications 

Months sustained 
by assets 

relative ranking 


Percentage in wealthiest group 
Percentage in middle group 
Percentage in poorest group 





from the enumerators. The poorest group 
constituted between 41.3 and 49.9 percent 
of the households. The lowest estimation 
was based on the asset coverage measure. 
This may be optimistic, however, because 
we did not control for differences in quality 
of assets or difficulties in disposing of as- 
sets in calculating asset values. The highest 
estimation is from farmers' own evaluation. 
While the analysts selected the cutoff, it is 
not surprising that a large number of people 
rank themselves below their neighbors in 
many categories. 

Finally, we compared the results across 
classification measure to see whether the 

households are similarly classified across 
measure. In fact, few households fall into 
the same category for all three different clas- 
sifications. However, the vast majority re- 
ceived the same score in at least two of the 
three classifications. We found, for instance, 
that 13.2 percent of the households were 
placed in the poorer group under all three 
classification measures. A further 28 percent 
were rated as poor in two of the three clas- 
sification measures. This creates a "poorer 
group" of 41.2 percent of households. At 
the other end of the spectrum, a small group 
of 6.6 percent of households was rated as 
wealthy in at least two of three measures. 


Household-Level Livelihood Strategies 
and Their Context 

In this chapter, we look at the main types of livelihood strategies that households pursue, 
linkages within and between strategies, and the strategies' implications for poverty reduc- 
tion. To try to understand these strategies fully, we also consider why people choose par- 
ticular livelihoods. In particular, we focus on the impact of these choices on people's life 
chances, household-level division of labor, notions of good farming, and outcomes. "Liveli- 
hoods" here refers to people's way of life. It covers what people do for their survival: striving 
to make a living, attempting to meet various consumption and economic necessities, coping 
with uncertainties, responding to new opportunities, and making a choice between different 
value positions (Long 2002). Therefore, in addition to finding shelter, transacting money, and 
preparing food to put on the family table, livelihoods are also about the management of rela- 
tionships, the affirmation of personal significance and group identity, and the interrelation of 
each of those tasks (Wallman 1984, 22). As such, livelihoods are also about the image people 
desire to project of themselves, and the value system informing this perceived identity (Omosa 
1998, 137; Omiti and Omosa 2002, 9). A person's livelihood is therefore an ongoing and dy- 
namic process. 8 

Types of Livelihood Strategies Pursued, How and Why 

Generally, rural households pursue several livelihood strategies, both on- and off-farm (Table 
5.1). In Siaya and Vihiga districts of western Kenya, most households interviewed pursued at 
least one of the following sources of livelihood: arable farming, livestock rearing, business, 
employment, and remittances. 

In the pilot areas, a majority of household members interviewed were students and there- 
fore dependent on other household members for their subsistence. The rest of the respondents 
were engaged in some productive work, with the majority being farmers or farmworkers. Only 
a negligible proportion of the respondents were reportedly engaged in off-farm employment, 
signaling the fact that agriculture is central to the livelihoods of most households in rural 
Kenya. Farming as a livelihood strategy entails the cultivation of crops or livestock rearing. 

8 The sections that follow focus mainly on occupations and income strategies, in part because of their importance 
to the study, and in part because of the experience of the fieldworkers who tended to be more oriented toward 
picking up these dimensions of livelihoods. However, there is also discussion that touches on identity, aspirations, 
recognition, social status, and so forth. 



Table 5.1 Livelihood strategies 
pursued by individuals in pilot villages 
(130 households) 

Livelihood strategy 









Farm help 



Home worker 



Casual non-farm labor 






Civil servant 






Skilled labor 






Source: Quantitative Survey 2002. 

Crops cultivated include cereals, legumes, 
root crops, and horticultural crops. The live- 
stock include cattle, goats, and sheep. 

One of the main features of the liveli- 
hood strategies pursued in rural Kenya is 
the fact that several strategies are applied, 
sometimes in combination, sometimes in 
succession, with the possibility of making 
reversals. The question therefore becomes: 
How are these strategies applied and under 
what circumstances? The various case study 
accounts suggest that generally, choices de- 
pend on the resources at hand, perceptions 
of incentives (rewards and costs), the desire 
to belong and fear of isolation, and how 
events unfold both for individuals and for 
their networks, wherever they are located. 

These criteria for choices are illustrated 
by an analysis of the livelihood strategies 
of Sufu, a 35-year-old farmer. As detailed 
in Box 5.1, Sufu started off as a small-scale 
farmer growing vegetables for sale. Within 
farming as a strategy, he preferred tomatoes 
to maize farming and worked out ways to 
improve his earnings. However, about two 
years later, he moved out of farming to take 
up casual employment secured through a 
brother. However, he also left that situation 
after a year because of the uncertain nature 
of casual employment. 

Nevertheless, he had saved funds, and in 
1991 opened a grocery shop using his sav- 

ings. This particular source of income trans- 
formed Sufu's life for the better, but the 
good fortune did not last. He was frequently 
attacked by armed robbers, and these set- 
backs reduced his capital base drastically. In 
1994, ICRAF came to his village, prompt- 
ing Sufu's return to farming. The same year, 
he married. Three years later, he moved into 
politics, leaving many of the farm activities 
to his wife. 

Sufu's account shows that the circum- 
stances that determine choices vary from 
time to time. For instance, after Sufu sat for 
his secondary school certificate examina- 
tions, he went into small-scale agriculture 
because it was not possible to continue with 
education. Thereafter, the choices he could 
make within agriculture now depended on 
his knowledge base and ability to seek in- 
formation. We therefore see him linking up 
with a farmer friend who was earning a fairly 
good living from tomato farming. But his 
friend earned far more from his tomatoes 
than Sufu did, because his farm was near 
a main highway and he employed superior 
crop husbandry practices. But we also see 
Sufu preferring cabbage to tomato farming 
because the former require fewer and less 
expensive inputs. In other words, in addi- 
tion to having a good market and access to 
required knowledge and skills, these liveli- 
hood choices also depend on one's capital 

On the other hand, people shift their 
livelihood strategies because others have 
suggested so or just for a change. Hence, in 
spite of having invested quite a bit in terms 
of knowledge and skills, Sufu easily moved 
out of farming to take up casual employ- 
ment. The availability of this employment 
was a result of the fact that he had a brother 
in town who found him this job. However, 
only a year later, Sufu returned to the village 
but with sufficient capital to invest in some 
off- farm employment in spite of having had 
some experience in agriculture. Yet, vulner- 
ability to armed attacks pushed Sufu out 
of this otherwise lucrative strategy and back 
to farming. In many ways, therefore, it is 


Box 5.1 Relay Type of Strategies 

"My name is Sufu and I am 35 years old. I completed secondary school education in 
1987, after which I spent two years at home engaged in small-scale agriculture. I planted 
onions, cabbages, sukuma, tomatoes, and beans. 

"I preferred planting the 'money maker' tomato variety because it used to fetch a 
relatively good price as compared to maize. I used only compost manure. Later on, my 
skills were much improved after taking lessons from a farmer friend from a neighboring 
village who was able to fetch 200 KSh per crate of tomatoes. This was mainly because 
my friend lives near the main road to Eldoret; he applied insecticides and was able to 
time well so that harvest time coincided with the dry season. This friend also taught me 
cabbage planting. At that time I preferred cabbage farming because it is less costly and 
it requires less labor. I was also able to sell all my cabbage to Emusire High School. 

"I spent my earnings on fashion trousers since at that time I was still a young man. 
I also paid school fees for my younger brother and bought some maize for my parents. 

"In 1990, 1 left for Nakuru at the invitation of my brother, who had found me a ca- 
sual job at the post office. I, however, quit this job later in the year, because I had been 
a casual for a long time. 

"In January 1991, 1 opened up a grocery shop in the village selling cabbage, sukuma, 
and retail goods. This time I had a lot of confidence because I was already experienced 
in business and I had some knowledge of accounts. I started with a stock of 10,000 KSh 
and at the end of that month the stock was 15,000 KSh. The capital for my business was 
from the savings of my former job. 

"This business was very successful and profitable. After six months I had already 
acquired four cows and six sewing machines. At the end of that year I built a semi- 
permanent house at a cost of 18,000 KSh. 

"In 1992, disaster struck: thugs attacked me and took away all the stock, valued at 
more than 25,000 KSh. I was forced to sell two cows to restock. In September of the 
same year, armed thugs robbed me of goods worth 67,000 KSh and this demoralized me. 
I closed the shop for one month, following which an Asian friend in Luanda gave me 
goods on credit. This time around I employed four men to transport and sell goods out- 
side the shop and to avoid too much stock. I also made burglarproof doors and we re- 
quested the local District Officer to put up a police camp at the Chief's office. However, 
the posting of police was delayed by lack of houses and an office because some mem- 
bers of the community were reluctant to contribute toward the construction of houses for 
fear of police harassment, especially because of local brew. 

"By 1993, my stock had fallen from 150,000 KSh to a mere 40,000 KSh, because 
multiparty politics introduced market-regulated policies while sources of income for our 
customers remained the same. Goods such as fruit juice, bread, and margarine became 
luxuries and were therefore never purchased. Furthermore, the Asians who used to give 
us credit stopped because they feared the outcome of the multiparty election and change 
of government. As my business fortunes dwindled and I was contemplating what next, 
ICRAF came to our village in 1994. This was an opportunity for me to rest from shaky 
business and venture into agriculture. The same year I decided to marry. I now have five 
children, two of whom I had sired before marrying. 

"In 1997, I joined active politics and left farm activities wholly to my wife. How- 
ever, our maize harvest dropped due to poor maintenance. I was so discouraged that the 


next season I decided to plant napier grass with the hope of getting a zero-grazing cow. 
This was because the few farmers that I knew who kept zero-grazing cows are able to 
pay fees and buy food and they could even afford a beer. I was so thirsty for these cows 
that at some point I wanted to sell all the indigenous cows I had, but my wife resisted. 
Once I get money, I am planning to concentrate on tomato farming, because I already 
have knowledge on how to use tithonia and manure as fertilizer." 

Source: Isikhuyu Village, Vihiga District. 

difficult for a single household to improve 
its level of development without comparable 
improvement by others. 

It is also apparent that ability to combine 
livelihood strategies is sometimes dependent 
on availability of labor. Sufu's account is a 
good example of a relay type of approach. 
While he was still a single man, Sufu tended 
to use only one strategy at a time, some of 
which had no direct linkage with succeed- 
ing ones. Although there is no mention of 
the reasons why he preferred such an ap- 
proach, there is a possibility that he faced 
labor constraints or could easily meet his 
needs. Hence, as soon as he married, Sufu 
started combining active politics with farm- 
ing, which, however, he "delegated" to his 
wife. Probably because he was now not as 
committed and the person in charge had not 
had as much exposure as Sufu regarding 
improved farming methods, their yields 
dropped. We therefore see Sufu making 
adjustments, some of which, however, are 
contested by his wife. 

Indeed, some of the other case studies 
also show that in an attempt to straddle, some 
livelihood strategies actually contradict and 
therefore interfere with the success of indi- 
vidual strategies. In several instances, how- 
ever, these livelihood strategies complement 
one another to the extent that many of them 
cannot be pursued in isolation. Among the 
issues that are therefore central to this im- 
pact assessment is the need to understand the 
driving force behind the choices that people 
make and why they sometimes remain in 
strategies that seem unprofitable. Some of 

the issues that could explain choice among 
rural households is people's notion of good 
farming and how this influences the type of 
strategies that they pursue to earn a liveli- 
hood, and the nature of investments that 
they put in place, including the soil fertility 
replenishing (SFR) technologies adopted. 
Therefore, whereas the rural poor may be in 
a position to appreciate the dynamic changes 
around them, they are often unable to take 
full advantage of opportunities that are 
perceived as capable of occasioning a turn- 
around. As such, whereas the rural poor may 
be aware of some of the opportunities that 
are available in the vicinity, such as good 
farming practices, they are often constrained 
from taking advantage of such opportunities. 
In other words, making opportunities for 
poverty reduction available does not neces- 
sarily mean that they will be accessible to 
the poor. In addition, failure of the poor or 
target group to take up these technologies 
does not imply their lack of knowledge of 
what is required to make their situation bet- 
ter. Poverty can therefore coexist with oppor- 
tunities meant to enable the poor to make 
their situation better for as long as these 
opportunities remain inaccessible. 

Local Notions of Farming 
as a Livelihood Strategy 

The question is, therefore, who is a good 
farmer? It is noted that there are different 
types of farmers and farming systems and 
these have changed over time. Local notions 
about good farming are based on people's 


Box 5.2 Types of Farmers and Farming Systems 

Maria is 46 years old and the mother of ten children, four of whom are deceased. Maria 
and her husband lived in Mombasa for over 20 years and returned to the village only in 
1993. At the beginning of 2001, her husband migrated to Busia to operate his newly 
opened retail shop. Prior to this, Maria's husband was quite idle, because he was never 
acquainted with farmwork. Other than having lived in town for many years, Maria's hus- 
band grew up among many girls and therefore never did much work during his youth. 

Maria works on the farm with her daughter, Grace, who although out of school, in- 
dicated that she goes to the farm only because she has no alternative. The other younger 
siblings help with farmwork only during the holidays and over the weekends, especially 
in peak working periods. They own about one acre of land and almost one half of this is 
occupied by the homestead. 

Since returning from Mombasa where she worked as house-help in a white man's 
house, Maria has been engaged in farmwork to raise food for her children. In 1997 she 
registered as a farmer with ICRAF and was taught how to use tithonia to improve land 
productivity. She decided to plant tithonia on alleys in her farm to cut down on the labor 
and time that is required to look for it from far away. However, she has used rock phos- 
phate only once, when ICRAF supplied them with some. 

Maria has never planted other fallow crops such as C. grahamiana and T. voghelli 
because she claims she does not have enough land to leave fallow for a whole year be- 
cause "I have to plant in the two seasons to get something to feed my children." She first 
learned about the use of tithonia from a neighbor and later attended village workshops 
organized by ICRAF. They were taught how to make compost manure using tithonia. 
For one year, Maria has not attended ICRAF meetings and workshops to gain knowl- 
edge about the new technologies. When I asked her why she does not attend these meet- 
ings, she said, "Jotelo mag ICRAF man e gweng' ka obuono jomoko" (The ICRAF 
agent/contact in the village is biased; he gives information to some people and not others). 
Because of this, many people are not as enthusiastic as before and they therefore lack 
current information. She added that "ma omiyo tinde ok ati maber" (because of this I do 
not work as well as I did then or as I could had I got the new technologies as taught by 
ICRAF staff). 

When she can afford it, Maria uses inorganic fertilizers. Her husband or her son helps 
her to buy the fertilizer (DAP). She states, however, that continuous use of inorganic fer- 
tilizers destroys the soils and that the fertilizers should be used together with compost 
manure. She says that the inorganic fertilizer is good only with plenty of rainfall; other- 
wise all the crops dry up. So, even when she has fertilizer, she uses it only during the 
long rains. 

When I asked her how she detects soil fertility levels, she said that she knows this 
by looking at the health of the plants growing on the farm, crop yields, and the kinds of 
weeds. "When I see Striga weed, then I know that the soil is poor." 

Maria plants only indigenous crop varieties because the hybrids are too expensive 
for her and many times in short supply. She also says that hybrid maize is not sweet for 
eating when green. 

Maria feels that her own piece of land is not enough and would like to hire more land 
for farming. She has once hired land for a year. This was situated far from her home and 
although the crop did well she did not harvest much because people stole most of it. This 


time she tried to get land closer to her home but it was not possible. She says she does 
not use tithonia on hired land because the owner may decide at any time to terminate her 
contract after she has improved the soils through the use of tithonia. "Some people have 
chased their tenants when they realize that the tenants are getting good yields. Therefore 
many people only use inorganic fertilizers on hired land." 

Source: Sarika Village, Siaya District. 

aspirations and these largely hinge on out- 
put and recognition from neighbors and 
friends, an aspect of people's livelihoods. In 
this section, we explore some of these issues 
with the aim of understanding further the 
various livelihood strategies that households 
pursue. In particular, we look at the different 
types of farming, what people perceive as 
good farming, and how these relate to soil 

Maria's account (Box 5.2) suggests that 
whereas most people are aware of what it 
takes to be regarded as a good farmer, many 
of them become constrained by circum- 
stances in their attempts to become good 
farmers. Further, in attempts to adapt to a 
diversity of situations, different types of 
farmers and farming systems emerge. 

Hence, although Maria's husband is 
present on the farm, he has decided not to 
get involved in farming and this therefore 
takes away from him the decision-making 
power and related choices. But, because of 
this, too, Maria cannot go for expensive farm 
inputs and she therefore suffers the same 
fate as many female -headed households. For 
instance, much as tithonia is more effective 
when used in combination with rock phos- 
phate, Maria used the latter only once when 
it was supplied to her free of charge from 
ICRAF Subsequently, she has been using 
only inorganic fertilizers whenever these are 
purchased for her. 

Other evidence, however, suggests that 
reluctance to use inorganic fertilizer goes 
beyond finances to include how people 
perceive the dangers associated with these 
technologies. On her part, Maria uses in- 

organic fertilizers only during the wet season 
when, in her view, they cannot destroy the 
soil. This therefore means that the choices 
that people make regarding farming activ- 
ities and that ultimately characterize their 
farming styles depend on how they per- 
ceive any of the practices that they engage 
in, irrespective of whether they are recom- 
mended or not. 

Furthermore, farming styles are also de- 
pendent on access to land. Hence, Maria is 
unable to adopt fallow crops because of the 
limitations arising from the size of her land. 
And, whereas she can afford to hire in addi- 
tional land, the expectations of those leasing 
out land make it difficult to implement soil 
fertility replenishing technologies and there- 
fore determine whether one actually takes 
up these technologies fully or not. In other 
words, no matter the level of exposure to 
soil fertility replenishing technologies, adop- 
tion as a strategy to enhance one's liveli- 
hood is also dependent on availability of 
basic inputs, land being one of them. There- 
fore, much as soil fertility may have been 
successfully identified as a prerequisite to 
good farming practices, actual implemen- 
tation is subject to vulnerabilities, such as 
people's failure to honor the contractual 
obligations governing leasing out land. 

Linkages between and 
within Strategies 

In light of the foregoing, are there differ- 
ent types of farmers? According to Maria's 
account above, there are different types of 
farmers, but their categorization depends less 


on their knowledge and skills than on avail- 
able resources — and how individuals per- 
ceive the opportunities and risks facing them. 
Are there any linkages therefore between 
and within the strategies that households 

Generally, the livelihood strategies that 
households pursue vary with gender to the 
extent that women diversify much more than 
men do. For instance, while Sufu seems to 
pursue only one strategy at a time, Maria 
alternates between several farm activities. 
However, these choices are also a function 
of resources. Generally, resource-poor 
households tend to diversify their strategies 
for fear of taking a risk or because none of 
the strategies can provide adequately on its 
own. Therefore, even though her land is 
small, Maria continues to combine several 
crops just to eke out a living. On the other 

hand, Sufu's strong asset base allows him 
the freedom to concentrate on one approach 
at a time. These combinations also depend 
on the nature of risks facing any one partic- 
ular individual and how he or she perceives 
available opportunities, if any. 

Last, all these strategies and livelihood 
approaches are linked since they largely 
constitute a people's identity and therefore 
provide a sense of belonging. Nearly all res- 
idents tend to follow the norms of behavior 
in their home area. For instance, in spite of 
its lack of rewards, subsistence farming 
has persisted — sometimes just because peo- 
ple wonder "what the neighbors will say" 
should they change. Hence, the struggle to 
belong and the continued search for identity 
forces some to continue actions that they 
might otherwise gladly put aside. 


Processes and Patterns of Adoption 

This chapter elaborates in detail what specific factors have influenced the adoption of 
soil fertility replenishment (SFR) technologies. 
First we detail which assets ICRAF thought would be vital for adoption, particu- 
larly with reference to the sustainable livelihoods (SL) framework. In the second section, we 
describe the actual patterns of adoption in the region, in both pilot and non-pilot areas, based 
on the quantitative data. At this level of analysis we will present a first typology of adopters, 
one that is developed and used by the disseminators themselves. The third section deals with 
decision-making by households regarding the adoption of agroforestry technologies. 

Explaining the exhibited patterns of adoption is an important task of this chapter. The most 
important issue is the extent to which the poor households, as defined in the previous chapter, 
are able to use improved fallow and biomass transfer technologies. In the analysis, two im- 
portant social dimensions deserve specific attention. We look at which assets play a key role 
and whether they have shaped the adoption of SFR-based practices. We also provide an ac- 
count of issues of control and mobilization of resources. Particular attention is given to the 
significance of land tenure arrangements. The analysis is disaggregated in terms of gender and 
age and other relevant forms of social differentiation. Furthermore, it elaborates the links be- 
tween the adoption and perception of the technologies, and the way these were introduced into 
the communities. Last, a social, typology of adopters is presented and elaborated. This typol- 
ogy differs from the first one in that it aims to capture local images and labeling. 

The fifth section draws our attention to how other technologies available to replenish soil 
fertility, such as fertilizer and cattle manure, are perceived and how they compare with agro- 
forestry-based technologies. The last section deals more specifically with how farmers adapted 
the original package to their benefit and the implications of the interventions for reducing 

Researcher Assumptions about Adoption of Agroforestry 

The premises underlying ICRAF's decision to invest in the development of agroforestry-based 
soil fertility replenishment systems can be summarized as follows: 

1. Agriculture is important in rural households' livelihoods, poor and wealthy alike. 

2. Reducing poverty in the short term requires improved incomes from agriculture. 

3. Increasing agricultural income will necessitate increased crop productivity. 

4. Soil fertility is critical for increased crop productivity. 

5. Fertilizer and animal manure is out of the reach of the poor. 

6. Households lack cash, but can spare some land and sufficient labor (especially at non-peak 
periods) for improved soil fertility management. 



7. Women will be empowered to plant 
trees for soil fertility inputs, even if 
they are not allowed to plant other 
types of trees. 

Given these assumptions, ICRAF, KARI, 
and KEFRI embarked on a collaborative 
research and development project in western 
Kenya. This impact-assessment research 
will allow for a quantitative validation of 
hypotheses 5, 6, and 7 — and some insight 
into hypothesis 4. For hypotheses 1, 2, and 
3, we rely on the qualitative analysis. 

Actual Patterns of Adoption 

In this section we present evidence of adop- 
tion of the improved fallows and biomass 
transfer agroforestry systems, in both pilot 
and non-pilot villages. For the former, all 
households were monitored, so the figures 
show the rates of adoption. For the non-pilot 
villages, our ultimate sample was stratified 
on the basis of use of the technologies, so 
they are not representative. For this sample, 
we show only how use has evolved over 
time. For estimating the rates of adoption in 
the non-pilot villages, we present the results 
of preliminary censuses from these sites con- 
ducted in order to develop sampling frames. 
Tables 6.1 and 6.2 and Figure 6.1 show 
the use of the agroforestry technologies over 
time in pilot and non-pilot villages. Distinc- 
tive patterns emerge inside and outside the 
pilot area. Inside the pilot villages, there 
was a rapid surge of use between 1997 and 
1999, when usage reached about one quar- 
ter of households for each technology. The 
year 2000 saw a significant decline in use, 
followed by a recovery in 2001. In 2001, 
15.2 percent of households were using im- 
proved fallows and 16.7 percent were using 
biomass transfer. A likely interpretation is 
that considerable technical support, along 
with the bandwagon effect, may have led to 

Table 6.1 Use of agroforestry in the 
pilot villages over time (as percentage 
of 1,538 households) 






1997 Long rains 



1997 Short rains 



1998 Long rains 



1998 Short rains 



1999 Long rains 



1999 Short rains 



2000 Long rains 



2000 Short rains 



2001 Long rains 



2001 Short rains 



n.a., data not available. 

early high rates of testing. This rise was fol- 
lowed by dis-adoption by those who did not 
receive sufficient benefits or were unable 
to manage the technology after ICRAF and 
partners reduced backstopping efforts (see 
later). Then in 2001, some early testers re- 
tried the systems and new testers surfaced. 
Outside the pilot villages, the dynamics 
differed considerably, with steady increases 
found over time for both technologies (and 
other SFR technologies as well). 9 Starting 
with just about 5 percent of households using 
agroforestry in 1997, the ending figures 
showed that 12.4 percent were using im- 
proved fallows and 21.6 percent were using 
biomass transfer in 2001. Biomass transfer 
therefore appears more desirable outside 
the pilot villages than inside. In both areas, 
however, there are new testers. 

There are further differences in the use 
of SFR within and outside the pilot villages. 
Within the pilot area, 54 percent of those 
who employ agroforestry are using both the 
improved fallows and biomass transfer, com- 
pared with only 38 percent outside the pilot 
area. Thus, when households have less con- 
tact with project staff, they more often than 

9 Note that while use of agroforestry was monitored annually within the pilot villages, the data for the non-pilot 
villages were based on recall from 200 1 . 


Table 6.2 Use of agroforestry in 
non-pilot villages over time (as 
percentage of 360 households) 

Table 6.3 Patterns of use of improved 
fallows and biomass transfer in the 
pilot villages (as percentage of 


Biomass transfer 

Improved fallow 

i,b9» nouseno 















Recent testers 






not prefer only one of the agroforestry sys- 
tems. This is further supported by the fact 
that of the new testers in the pilot villages, 
those who started after project staff had 
left, 88 percent are trying just one of the 

For the pilot villages, the data presented 
in Table 6.1 were analyzed to classify house- 
holds into different categories of adoption. 
The adoption dynamics were summarized 
into four mutually exclusive outcomes for 
each technology: 

1. Households that never used the tech- 
nology (non-adopters) 

2. Households that used the technology 
early on but never again (dis-adopters) 

3. Households that did not use the tech- 
nology early on but used it recently 
(recent testers) 

Figure 6.1 Adoption patterns of improved 
fallows and biomass transfer in the pilot 
villages overtime, 1997-2001 (as 
percentage of 1,630 households) 

Percentage of households 






1997 1998 1999 2000 2001 

t S 











Biomass transfer 

4. Households that used the technology 
throughout the period (adopters) 

Because the actual patterns of use varied 
a great deal, the rules for placing house- 
holds in any of the above categories are quite 
lengthy. For the most part, adopters will have 
used fallows more often than dis-adopters 
and testers, but some dis-adopters may have 
used fallows two times (that is, in 1997 and 
1998). Table 6.3 and Figure 6.2 show that the 
majority of households — about 60 percent 
— had not tried either of the technologies by 
2001. A greater percentage of households 
have adopted improved fallows (22.0 per- 
cent) than biomass transfer (15.0 percent). 
However, about twice as many households 
have recently tested biomass transfer than 
have tried improved fallows (14.6 percent 
to 7.6 percent). For both technologies, about 
10 percent of households tried and then 
dropped the practice. 

Censuses were performed at six differ- 
ent sites outside the pilot villages but nearby 
(about 1,000 households in all). Because the 
rates of use are expected to be relatively 
high compared to other non-pilot villages 
(indeed, this is one of the variables for strat- 
ification), these should not be taken to be 
representative. Table 6.4 shows that rates of 
use are very high in five of the six sites, 
ranging from about 24 to 59 percent. This is 
encouraging, given that technical support 
from the project in these sites has been rela- 
tively low. In fact, the site with the highest 
adoption rate (West Kanyaluo) is the one 
that received the least amount of attention 


Figure 6.2 Adoption patterns of improved fallows and biomass transfer in the pilot 
villages by 2001 (as percentage of 1,630 households) 

Percentage of households 




40 - 



10 - 


Improved fallows 
Biomass transfer 

I L 



Recent testers 


from ICRAF or any other intermediary. An 
umbrella NGO works in the Bukhalahire 
site and has assisted farmers there. Muhande 
is a former CARE village and along with 
Mwitubi hosts ICRAF technicians from 
time to time because of its proximity to the 
research center in Maseno. Shinyalu hosts 
researchers from KARI-Kakamega. Finally, 
Central Gem is a site where the main con- 

tacts for improved fallow dissemination were 
the members of a soil- and water-catchment 
committee, formed by extension workers. 

The use of improved fallows and bio- 
mass transfer has increased over time in the 
non-pilot villages. Overall use is increasing 
for both technologies, in contrast to the pilot 
villages, where use patterns seemed to mir- 
ror the intensity of ICRAF technical support. 

Table 6.4 Rates of use of improved fallows in early non-pilot area villages 

Site Number of households Percentage of households with improved fallows 

Bukhalahire (Busia) 
West Kanyaluo (Rachuonyo) 
Shinyalu (Kakamega) 
Mwitubi (Vihiga) 
Muhande-Arude (Siaya) 
Central Gem (Siaya) 




Note: Ugunja was enumerated, but a proper census with a full listing of households was not performed there. 


Table 6.5 Size of fallows over time in 
pilot villages (square meters) 



Size in 1998 



Size in 1999 



Size in 2000 



Size in 2001 



Change over 




"Calculated only for those farmers with at least two 
fallows occurring at least two years apart (2001 minus 
1998 size, or 2001 minus 1999 size, or 2000 minus 
1998 size). 

Table 6.5 shows that average fallow area was 
highest in 1998, dropping to a low in 1999 
and recovering somewhat in 2000 and 2001 . 
Fallow size was reduced in 1999, partly be- 
cause of both rainfall and seed supply con- 
straints in addition to farmer preferences. In 
2001, the mean fallow size was 440 meters 
squared or 0.04 hectares. While this does 
not sound like much, it should be recog- 
nized that the average farm size for many is 
about 0.6 hectares, of which perhaps 0.3 to 
0.4 is under maize. Further, the fallow sys- 
tem calls for a rotation of a fallow followed 
by three seasons of maize. If this pattern is 
followed, one would expect only one fourth 
of the maize area to be under fallow at one 
time — this would be between 0.075 and 
0.100 hectare. Viewed in this way, adoption 
intensity among those using fallows appears 
to be more significant. 

Table 6.6 shows the percentage of house- 
holds who planted tithonia on their farm. 
This is perceived as increased investment 
in the biomass transfer system by increasing 
the amounts of organic material available or 
reducing the labor required for collection 
of the material off-farm (or both). It can be 
seen that considerable planting occurred in 
1998, 1999, and 2001, over 11 percent of 
households in each case. It is not known 
why investment declined so significantly in 
2000. Continued monitoring will help to 
reveal whether this was just an anomaly or 
an early sign of saturation of interest. 

Table 6.6 Planting of tithonia biomass 
transfer systems on farm over time in 
pilot villages (as percentage of 
households planting) 

Year Percentage of households 





The issue of who decides to adopt (or not 
to adopt) SFR technologies is a tricky one. 
Customarily, among both the Luo and 
Luhya, it is the husband who makes such 
decisions. A researcher asking questions 
about intrahousehold decision-making tends 
to get the same answer ("The men decide"), 
even from women ("My husband decides"). 
In cases where the man in the household or 
compound is present, he is the recognized 
"owner" of the land and the head of the 
household, and men often decide to try im- 
proved fallows or green manure. This does 
not necessarily mean, however, that women 
have no say in such matters. As described in 
Chapter 3, one of the female farmers was 
the first in the household to receive knowl- 
edge about use of tithonia and the fallow 
trees; she shared this information with her 
husband after she had planted them on her 
plot. Another female farmer works alone 
because the husband has no interest in 
working on the farm, so she decides what 
kind of technology to employ on the farm. 
In the case of two wives of a Luo husband 
from the same homestead, one decided to 
adopt the SFR technologies while the other 
has not. Each has their own reasons. 

The situation as explained in Chapters 3 
and 5 is such that men often seek off-farm 
employment, leaving their wife or wives 
behind to cultivate the fields. In such cases, 
it is not unusual to hear women say that they 
have made the decision to experiment and 
plant fallow and green manure species. Yet 


women rarely mention that they decide what 
to do but rather give their husbands the 

Some of the case studies show that the 
decision to adopt or not has caused dis- 
agreements. In one particular case, both 
husband and wife pursue different farming 
practices, as a preference. In this case, the 
man first learned of the new SFR technolo- 
gies. But because he is an alcoholic with 
low social status in the community, his wife 
was not convinced that his new farm prac- 
tices were anything to emulate. Instead, she 
viewed them as a likely continuation of his 
wayward tendencies. 

One difference occurs at the level of pilot 
versus non-pilot villages. In the pilot vil- 
lages, women are active adopters. However, 
in the non-pilot villages, the few adopters 
are mostly men. 10 The reason could be that 
new knowledge has to be searched for from 
a distance as compared to the pilot villages, 
where the knowledge is brought close to 
home and women are able to attend the 
learning sessions. 

For several major reasons, men have an 
advantage over women in accessing infor- 
mation. Relatively free of household chores, 
men have a great deal of time to be away 
from home. Because of their larger social 
space, they can also interact with other peo- 
ple more freely, and go for exchange visits 
and other meetings without being questioned 
as to their whereabouts. Men's absence 
from home also does not interfere with 
domestic work such as childcare, cleaning, 
cooking, or fetching water and firewood 
to keep the household going. One female 
farmer in a pilot village, an elite woman, ed- 
ucated to form four, said that a hindrance to 
women's participation in exchange visits 
is the allegation that some of them befriend 
the ICRAF staff. Therefore any trip is sus- 
pect as a potentially illicit opportunity. 

Higher adoption of the SFR technolo- 
gies by women in the pilot village was in- 
spired mainly by the fact that they needed 
higher yields to be able to provide food for 
their households. One woman said, "We 
work hard and do what we can to increase 
our yields, because we are the ones who 
stay with the children at home. If we can- 
not give them food, they will cry 'on our 
heads.' " 

Education was not found to play a major 
role in the decision of farmers to take up 
technologies. Less educated women excel 
in adopting new technologies as long as 
explanations are provided in simple terms. 
Interestingly, the data clearly show that 
women understand the management and 
impacts of SFR technologies better then 
men. This may help explain why women are 
enthusiastic about SFR and decide to adopt 
such techniques. 

Because many of the technologies were 
taught in a practical manner, it was easy for 
the farmers to try them. Education was found 
to be important, however, when it came to 
understanding technical issues such as the 
science of how the fallows and tithonia 
work in the soil, including the release of 
nitrogen gas and uptake by crops. 

Gender must also be considered as 
women must ask their husbands' permission 
to attend seminars and meetings. When the 
bus broke down during a study tour, the 
reactions showed how seriously the gender 
roles and constraints must be taken. When 
the bus broke down, ICRAF was forced to 
provide hotel accommodations for the par- 
ticipants, including the women. The hus- 
bands, it turned out, were not happy with 
the fact that their wives had spent the night 
away from home. While such sentiments 
may be part of gender relationships in gen- 
eral, it remains unclear whether the angry re- 
actions might also have something to do with 

10 As will be shown in a later section, the rate of adoption among female heads is similar to that of males in both 
the pilot and non-pilot villages. These perceptions by community members on gender participation are more 
likely related to a comparison of absolute numbers of male and female users. 


ICRAF and/or the way SFR technologies 
had been introduced. 

Explaining Adoption and 

This section reviews the analysis of the 
quantitative and qualitative data. 

Quantitative Findings 

Pilot Villages. A multinomial logit regres- 
sion was run to examine the effect of several 
explanatory variables on the likelihood of 
being a dis-adopter, a tester, or an adopter, 
relative to having never tried the technology 
(an improved fallow or biomass transfer, 
examined in separate models). The distribu- 
tion of these outcomes for each technology 
has already been presented in Table 6.3. 
While some categories have a relatively low 
percentage of households, the large number 
of observations permits us to include them 
in the analysis. 

The explanatory variables are from a 
survey conducted in 1997 in 17 villages. 
Thus, all the variables included in the model 
are predetermined in relation to the adop- 
tion variable. Moreover, most of the vari- 
ables could be treated more rigidly as ex- 
ogenous. Such exogenous variables include 
village location in pilot area or not (selected 
by external organizations), ethnicity, educa- 
tion level of household head, age of house- 
hold head, and owned land area. There may 
be some selectivity of households over the 
number of adults in the household as well 
as the decision-making structure. Mainly, 
it may be that other exogenous variables 
(e.g., owned land area) may affect migration 
decisions of households and therefore the 

probability of having a female head with the 
husband living away. Also, the wealth index, 
based on farm management practices (e.g., 
ability to hire labor) and assets, is likely to 
be related to similar exogenous variables. ' ' 
Different models were thus run with and 
without these choice variables, as well as 
with other non-exogenous variables. The 
model reported in the table is a compromise 
that attempts to include as much as possible 
only exogenous variables, but that also in- 
cludes other variables highly relevant to 
poverty, the main focus of the research. 

Table 6.7 shows the results of a multi- 
nomial logit analysis of dynamic use patterns 
of improved fallows in the pilot project 
areas between 1997 and 2001. In general, 
the included variables appear to be very 
important in distinguishing between dis- 
adopters and non-adopters, to some extent 
between adopters and non-adopters, but not 
very relevant to distinguishing between re- 
cent testers and non-adopters. Rather than 
describing results outcome by outcome, we 
shall instead analyze by variable across the 
different outcomes. 

First, we shall discuss the variables most 
closely linked to poverty, the wealth index, 
the type of household, and farm size. The 
wealth index was not statistically significant 
in any of the pairwise comparisons, suggest- 
ing that the different use patterns are neutral 
with respect to wealth — the poor are as 
likely to be adopting as the wealthy. House- 
hold type was also not related to adopting 
improved fallows — the technology is being 
adopted to the same extent by female- 
headed and other nontraditional household 
structures as by the more common male- 
headed, monogamous household. A final 
variable linked to poverty, farm size, shows 

"Note that farm size is just one component of the wealth index, as many other variables were cited as essential 
indicators or contributors to wealth. We use the terms "wealth category," "wealthy," or "poor" to relate to the 
classifications based on the broader wealth index measures. It should be recognized, however, that all the "near- 
landless" households are not included in our group of "poor" households. Thus, in the conclusions chapter, we 
reinforce this difference by using the phrase "non-land wealth" to reflect the findings related to the wealth index. 
We note separately the results of analyses of the effects of farm size on adoption and impact. 


Table 6.7 Household factors related to adoption of improved fallows in pilot villages, 
1997-2001 (/i= 1,583) 


Used early 




and dropped 

recently only 

throughout period 








Pilot village 







Luo household 







Number of adults 







Female head — husband away 







Female head — no husband 







Male head — polygamous or single 







Secondary education 







Upper primary education 







Lower primary education 














Owned land area 







Wealth index 







Percentage of cases observed 




Notes: Omitted outcome is the group of farmers never trying improved fallow, /^-values in parentheses; "sig- 
nificant at at least 5 percent level; '"significant at 10 percent level. 

a different pattern. Non-adopters of fallows 
have smaller farm sizes than dis-adopters 
and adopters. Somewhat encouraging is that 
households that are newly trying improved 
fallows tend to have farm sizes indistinguish- 
able in size from those of non-adopters. 
Using the land/adult labor ratio in an alter- 
native regression, it is found that greater 
ratios are positively related to the adoption 
of fallows (though not significant for dis- 
adoption or recent testing). Thus, for adop- 
tion, land is a more important household 
constraint than is labor. 

Among other variables, being in one of 
the focal pilot villages (10 of 17 villages in 
the pilot area) was instrumental in testing 

fallows at an early date, whether the practice 
was continued or not. However, location is 
not important for recent testers — this is 
suggestive that recent testing is related less 
to technical backstopping, other external 
motivations, and to the spillover effect of 
larger numbers of nearby users. One inter- 
pretation is that because fallows and their 
effects are highly visible, many farmers were 
able to make early decisions about whether 
to test them (hence the relatively few recent 
testers) and thus there are few relationships 
between explanatory variables and recent 
testing. Early use was similarly higher 
among Luos as compared to Luhyas. How- 
ever, just like the case with the pilot location 


Table 6.8 Household factors related to adoption of biomass transfer in pilot villages, 
1997-2001 (/i= 1,583) 


Used early 




and dropped 

recently only 

throughout period 








Pilot village 







Luo household 







Number of adults 







Female head — husband away 







Female head — no husband 







Male head — polygamous or single 







Secondary education 







Upper primary education 







Lower primary education 














Owned land area 







Wealth index 







Percentage of cases observed 




Notes: Omitted outcome is the group of farmers never trying biomass transfer, p-values in parentheses; 
nificant at at 5 percent level or less; *significant at between 5 and 10 percent levels. 


variable, new testers are equally likely to be 
Luhyas as Luos. 

Education level and age of the house- 
hold head were not related to adoption of 
improved fallows (or to early testers). Thus, 
those households using fallows in 2000-01 
are similar in terms of household-head char- 
acteristics as households who never tried 
fallows. Older household heads and those 
with a secondary education were less likely 
to have dis-adopted fallows rather than hav- 
ing never used one. In other words, younger 
household heads were more likely to be ad- 
venturous and try a fallow but then abandon 
it than an older household head who had 
never tried a fallow. 

For biomass transfer, the included ex- 
planatory variables distinguished the three 
different outcomes from the base case of 
non-adoption about equally well (see Table 
6.8). Several variables were important for 
each pairwise comparison. The wealth index 
variable was not related to adoption of bio- 
mass transfer compared to non-adopters 
(thus wealth is not linked to the adoption of 
either agroforestry practice). However, the 
more wealthy households are more likely to 
have dis-adopted or recently tried biomass 
transfer. This means that should the recent 
testers become adopters, the profile of 
adopters will become wealthier. The struc- 
ture of household is not related at all to the 


pattern of use of biomass transfer, so that 
the technology is completely neutral with 
respect to household decision-making struc- 
tures. The size of the farm is positively re- 
lated to the adoption of biomass transfer, 
though not to decisions to dis-adopt or test 
in recent times. However, the supply of labor 
is also very important in the use of biomass 
transfer (all three outcomes where use has 
taken place). When the land/labor ratio is 
used as a regressor (rather than the two vari- 
ables independently), it is not significantly 
related to any of the outcomes, implying 
that neither land nor labor dominates the 
other as a constraint. 

A Luo household in a focal pilot village 
is much more likely to have adopted bio- 
mass transfer than a Luhya household in a 
nonfocal village. Moreover, new testers are 
similarly likely to be drawn from this pop- 
ulation. Because external assistance has 
largely been withdrawn from these sites, 
this result likely indicates that there has 
been significant farmer-to-farmer learning 
in which large concentrations of early users 
lead to large concentrations of new testers. 
The reason for higher use of biomass among 
the Luo is not clearly known. One hypothe- 
sis is that their strong subclan affiliation 
may lead to increased use among clusters 
of households. But we find only partial sup- 
port for this, with very high or low rates of 
adoption in about half the Luo villages, but 
moderate levels in the other half. 

Education and age play a stronger role 
in use of biomass transfer than they do for 
improved fallows. More educated household 
heads are more likely to have adopted bio- 
mass transfer than noneducated household 
heads. Similarly, there is some evidence that 
more education leads to less dis-adoption 
than non-adoption. Age of household head 
is not statistically related to adoption, but 
younger heads are more likely to be recent 
testers as well as being dis-adopters than 
those who had never tried biomass transfer. 
So younger household heads seem to show 
great interest in biomass transfer, but have 

not always had sustained interest or ability 
to maintain the use of the practice. 

Non-Pilot Villages. A multinomial logit 
model is used to analyze adoption behavior 
of households in non-pilot villages, as was 
the case for the pilot villages. In this analy- 
sis, we again classified households into dif- 
ferent categories of technology use. One key 
difference, however, is that we did not know 
at what point a given household was ex- 
posed to information about the agroforestry 
practices (whereas almost all households 
were informed about the practices at the 
same time in the pilot villages). Therefore, it 
becomes harder to differentiate between 
testers and adopters. Further, the more re- 
stricted number of households (361) reduces 
the flexibility in number of distinct cate- 
gories. We therefore created three categories 
of households: (1) nonusers/dis-adopters, 
(2) infrequent users, and (3) frequent users. 
In the case of improved fallows, 15.8 per- 
cent were frequent users and 13.6 percent 
were infrequent users. The comparable fig- 
ures for biomass transfer were 18.0 percent 
and 19.1 percent. 

We used the same household explanatory 
variables as in the case of the pilot villages 
with the following exceptions. For house- 
hold type, all female-headed households 
were combined into a single dummy vari- 
able owing to insufficient numbers in sev- 
eral more disaggregated categories. Second, 
we reduced the number of variables depict- 
ing the education level of the household 
head to include primary and secondary/ 
above (as opposed to further splitting the 
primary education variable as was done in 
the pilot villages). Third, for wealth we ac- 
tually have more varied and rigorous mea- 
sures and include three alternative speci- 
fications in our model. Last, because the 
non-pilot villages cover a wide geographi- 
cal area, we include location dummies for 
each site. 

The results for the improved fallow and 
biomass transfer regressions are given in 


Table 6.9 Multinomial logit results for adoption of improved fallows in 
non-pilot villages (/i = 361) 




Female-headed household 

Polygamous male-headed household 

Primary education of head 

Secondary or greater education of head 

Age of household head 

Number of household members 

Farm size 

Wealth — Log of assets 

Wealth — Farmer-generated index of wealth indicators 

Wealth — Enumerator-generated middle wealth level 

Wealth — Enumerator-generated high wealth level 

Notes: The three alternative wealth specifications are tested in separate models. Explanatory vari- 
ables reported are for the Wealth — Log of assets specification. Where the results of the 
non-wealth variables change across specification, it is noted in the text. The two reported 
columns are to be compared to the omitted outcome of never having used the technology, 
p-values in parentheses; "significant at 5 percent level or less; *significant at between 5 
and 1 percent levels. 
a Eight location variables not reported. 

























































Tables 6.9 and 6.10. One key result is that 
there are hardly any statistically significant 
results among the household variables, con- 
trasting the results from the pilot villages. 12 
One statistical reason why this may be ex- 
pected is that the number of observations is 
about 20 percent of those in the pilot vil- 
lages and standard errors of estimates will 
be higher, all else equal. There is also greater 

geographical dispersion among the six non- 
pilot villages (in five districts) than in the 
17 pilot villages (in two adjacent districts) 
and therefore unobserved factors may have 
played a stronger role. 

The only household variable that was 
linked to the frequent use of improved 
fallows was one of the wealth variables 
(farmer perception of relative wealth), in 

12 There were many significant results among the location dummies. 


Table 6.10 Multinomial logit results for adoption of biomass transfer in 
non-pilot villages (/i = 361) 




Female-headed household 

Polygamous male-headed household 

Primary education of head 

Secondary or greater education of head 

Age of household head 

Number of household members 

Farm size 

Wealth — Log of assets 

Wealth — Farmer-generated index of wealth indicators 

Wealth — Enumerator-generated middle wealth level 

Wealth — Enumerator generated high wealth level 

Note: The three alternative wealth specifications are tested in separate models. Explanatory vari- 
ables reported are for the Wealth — Log of assets specification. Where the results of the non- 
wealth variables change across specification, it is noted in the text. The two reported 
columns are to be compared to the omitted outcome of never having used the technology, 
p-values in parentheses; "significant at 5 percent level or less; '"significant at between 5 
and 1 percent levels. 
a Eight location variables not reported. 

























































which case the more wealthy households 
were more likely to be frequent users as op- 
posed to non-using households. The same 
variable was positively related to infrequent 
use and the enumerator evaluation of house- 
hold wealth was also positively related to 
infrequent use. So although not all the 
wealth variables are producing similar re- 
sults, there are indications that wealth is 
important in the use of improved fallows. 
The only other significant result in the fal- 
low regression is that Luhya households 

were much more likely to be infrequent 
users as opposed to the Luo. This is difficult 
to interpret because the same variable has 
almost no influence whatsoever in regards 
to frequent use of improved fallows. 

A similar pattern emerges for biomass 
transfer. Only the wealth variable(s) is re- 
lated to the use of biomass transfer. In par- 
ticular, the asset and farmer measures are 
positively related to frequent use of biomass 
transfer. The farmer measure is also related 
to infrequent use and the enumerator evalu- 


ation of wealth is weakly positively related 
to infrequent use. So, a similar conclusion is 
that there is a positive link between wealth 
and the use of biomass transfer. No other 
household variables were statistically sig- 
nificant in the regressions. When the wealth 
variables are omitted altogether, the only 
change in statistical significance is with the 
labor variable in the biomass transfer re- 
gression, which now becomes significantly 
positively related to frequent use. 

There is a positive link between wealth 
and the uptake of the technologies in con- 
trast to the findings in the pilot villages. 
This may reflect the extra attention given to 
reaching the disadvantaged groups within 
the pilot villages, or could also be partly 
attributable to different measurements of 
wealth in the two sets of regressions. It is 
equally interesting to note that while farm 
size and labor constraints were apparent in 
reducing the uptake of improved fallows 
and biomass transfer in the pilot villages, 
such constraints did not emerge in the non- 
pilot areas. There is a marginally positive 
effect of labor on biomass transfer in the 
non-pilot areas, but the impact of farm size 
is almost nil. This issue requires further 

Analysis of Qualitative Data 

Qualitative data analysis confirms the results 
of the previous section, and also reveals fur- 
ther important issues and insights. Land, 
and particularly the size of a field, is an 
important factor explaining adoption and 
non- and dis-adoption. In Sarika, for in- 
stance, more than half of the farmers have 
an average field size of one acre, including 
the homestead. Many of the homesteads 
occupied 0.5 acre or slightly more. Those 
who could afford to rent land from others — 
mainly from older women — could escape 
the limits imposed by their own field size. 
The so-called "rich" in the community rent 
fields from those who are not able to work 
on their land. Yet for many, the cost of hir- 
ing land has become too high. It is in this 
context that one needs to understand that 

tithonia for biomass transfer purposes is 
favored compared to planting trees for im- 
proved fallows. An advantage of planting 
tithonia is that it can be planted or harvested 
anywhere — in hedges, on roadsides, and so 
on — and does not necessarily require extra 
space. In constrained acreage, improved 
fallows occupy space that normally would 
be planted with crops planted in the short 
rains (maize, cassava, and sweet potatoes) 
that play a major role in food security dur- 
ing difficult times. 

Labor, in terms of both quality and 
quantity, is key as well. Labor shortages can 
occur when able-bodied men migrate to 
off-farm jobs. Age also plays a role. Many 
dis-adopters argued that they were simply 
too old to apply the required management 
practices. Many of the people older than 60 
years of age, in fact, were not enthusiastic 
about the technologies. Even those who felt 
that they were good technologies that result 
in crop-yield improvement said that they 
would have done better with them as young, 
energetic people. Elders said that they no 
longer have enough strength for heavy 
farmwork. Some have already apportioned 
their parcels to their adult children. 

Many farmers said that harvesting, trans- 
porting, chopping, and eventually applying 
biomass is too labor intensive. While some 
refer to age, others, especially women, said 
their domestic chores and other livelihood 
activities such as small-scale businesses do 
not allow them time to use the agroforestry 
technologies. Some of the farmers have 
tried to modify the technologies — especially 
in the case of tithonia — to minimize labor. 
They skip certain steps; for instance, instead 
of mixing tithonia with soil, they may sim- 
ply broadcast the tithonia leaves (see, also, 
"Adaptations" later). 

A crucial issue that the quantitative 
research did not disclose is that labor is 
complicated not only because of availabil- 
ity, but also by issues such as control over 
labor. Labor may be available — but beyond 
the control of farmers. Control over labor 
is partly regulated by the complexities of 


kinship relationships (see Hebinck and 
Mango 2001; Mango 2002) and has partly 
to do with the history of Luhya working as 
"slaves" for the Luo. 

Adoption and Method of 
Technology Introduction 

Whereas the previous section attempted to 
explain processes of adoption with refer- 
ence to material assets (land, labor, financial 
capital, and so on), this section explores 
adoption from the point of view of social 
relationships between villagers and interven- 
ers. 13 The analysis of the data hints that the 
perception of the AF- based technologies — 
and the actual use, adaptation, or dis-adoption 
— is associated with both the social fabric of 
the communities and the social processes 
involved when these technologies were in- 
troduced. Looking at the available qualitative 
data, we must account for: 

The interfaces emerging out of relationships 
between villagers and encounters with 
representatives of intervening agencies 
such as ICRAF, KEFRI, and KARI 
The initial role of the seed market 
The roles of knowledge and its exchange 

Interfaces and Encounters: The ICRAF 
Agents and Favoritism. Generally, partic- 
ipants expressed a great deal of appreciation 
for the work of the Maseno station staff. 
But mixed feelings also persist. Interviewees 
said they "love the Maseno people" for their 
inputs, but do not like the "agents." By 
"agents," or more specifically "ICRAF 
agents," is meant certain individuals who 
gained a great deal of attention from 
ICRAF/KEFRI/KARI in their endeavors to 
introduce and disseminate agroforestry 
technologies in the region. Getting a great 
deal of attention means receiving a lot of 
visitors (VIPs such as the minister of agri- 

culture, district commissioner, director 
general of ICRAF, foreigners, extension 
workers) or sometimes even favors such as 
gifts. 14 A specific concern raised in the case 
studies is that these individuals were not 
democratically elected by the villagers, but 
hand-picked by ICRAF staff. 

A number of farmers in the pilot and 
non-pilot villages felt that the agents were 
biased in many of their interactions with 
farmers. They were said to give information 
to a few individuals and not others. This 
caused some of the farmers to cease attend- 
ing the meetings and workshops organized 
by ICRAF staff in the village. It was said, 
"they (the agents) only give information to 
whomever they please. Therefore it is not 
possible to attend a meeting for which you 
have not been invited." Another farmer said, 
"When it comes to exchange visits, he does 
not use the list he is given from the office in 
Maseno. He only takes his friends. And that 
he (the agent) does not want farmers to talk 
to ICRAF staff from Maseno without pass- 
ing through him." 

People frequently mentioned that partic- 
ular farmers were taken for field visits and 
others left out despite the fact that all of 
them were adopters. Those left out feel the 
technology is only for particular people. 
They blame the ICRAF staff for heavily re- 
lying on the agents to choose people who 
attend seminars and workshops. The agents 
tend to choose their relatives and friends. 
This issue of "agents" can partly be ex- 
plained with reference to jealousy. But it is 
widespread: an overview of the cases shows 
this concern mentioned in at least 40 per- 
cent out of a total of 40. Contentious focus 
group discussions also support this observa- 
tion (see Chapter 9). The "agent" issue also 
raises the question of how individuals 
manage to maneuver themselves into strate- 
gic positions. In both the Luo and Luhya 

13 For an elaboration of this notion, see Long (2002). 

14 This is also supported by Mango (2002). See, particularly, Chapter 8. 


villages, "danism" and political party affil- 
iations appear to play an important role. 
Furthermore, the social science literature 
provides evidence that this is not limited to 
Kenya and gives some clues on how dissem- 
ination of technologies through targeting in- 
deed helps empower certain individuals 
rather than communities. 15 Visits of ICRAF 
and other agencies clearly lend prestige and 
thus a fair amount of social capital. 

Another expressed concern is that 
"Wazungu [white people] have taken our 
land." In fact, scientists rent local plots for 
experiments and trials. That the local people 
themselves might not be part of the experi- 
ment or trial (thus feel no ownership) may 
be responsible for this concern. Not only 
might this lack of ownership encourage dis- 
adoption; it also reinforces the importance 
of social relationships with and between 
villagers, and between villagers and imple- 
menting agencies. In addition, interviewees 
mentioned a number of cases of theft of 
seeds or use of "magic." 

Ownership of trees is another frequently 
cited issue. Often farmers view trees as 
"CARE trees" or "ICRAF trees" rather than 
their own. This image is probably associ- 
ated with the way agroforestry technologies 
were originally introduced. Villagers ex- 
plained in one of the group discussions that 
ICRAF paid people to prepare the land and 
to plant trees for them in a trial held in the 
village. Agroforestry was clearly labeled a 
"Mzungu [white] thing." 

These comments suggest that we need 
to interpret such views, as well as favoritism, 
as part of the nature of social relationships 
that emerge over time between institutions 
such as ICRAF and individual farmers and 
communities. It is important to stress that 
such images (irrespective of being "right" 
or "wrong") and favoritism have negatively 
shaped people's perceptions of agroforestry- 
based technologies. This does not, however, 
say anything about the technology per se. 

The Seed Market. The SFR project gener- 
ated income opportunities for farmers, par- 
ticularly in the beginning when seeds could 
be sold at high prices. Quite a few people in 
the pilot villages took advantage of this sit- 
uation and made a good deal of money from 
the early seed market. Some of them even 
managed to buy a dairy cow, a perfect source 
of milk and manure (see Mango 2002 for 
similar cases). Some farmers decided to 
adopt after ICRAF promised to purchase the 
seeds; others adopted with the hope that 
whites who came to the demonstration site 
would give them money and farm tools. 
When this did not take place, dis-adoption 
began, owing to non-fulfillment of some 
farmers' expectations. 

The early situation of good markets and 
high prices for seed certainly colored peo- 
ple's understanding and perception of SFR 
and agroforestry-based technologies. This 
situation has, however, changed over the last 
few years. When seeds were no longer in 
demand by ICRAF and other agencies, agro- 
forestry and the selling of seeds in particu- 
lar lost their status for some people. Some 
of the seed sellers dis-adopted after prices 
dropped substantially. This process acceler- 
ated after 1999, when ICRAF was accused 
of abandoning the people in the pilot vil- 
lages. From ICRAF's point of view, such 
accusations are unavoidable and "part of the 
scaling up of our operations, which implies 
that resources had to be withdrawn from the 
pilot areas." 

Knowledge as a Resource. In the pilot vil- 
lages included in the case study research, the 
main source of knowledge regarding the new 
technologies is ICRAF. Although ICRAF, 
KEFRI, and KARI staff from Maseno work 
together, KEFRI and KARI have a low pro- 
file, as all visible vehicles bear the ICRAF 
logo. Therefore, for the farmers, a mention 
of Maseno means ICRAF. Only one farmer 
in Sarika knew the difference. 

5 See, with reference to cases in Mexico, Arce (1993) and Villarreal (1994). 


Although ICRAF may dominate the dis- 
semination of knowledge and skills, farmer- 
to-farmer sharing appears to be a close 
second. Particularly, seeing one's neighbors 
"doing it" emerges as crucial. Curiosity and 
being connected to their neighbors are im- 
portant elements in the dissemination pro- 
cess. Yet despite this and the fact that in 
most of the villages, people are kin in one 
way or another, there is some evidence that 
farmer-to-farmer extension has generated 
intense jealousy. Questions that remain 
unanswered include: Has this always or 
often been the case? Or is it specific to these 
interventions? Or might it be part and parcel 
of the dissemination approach? Is the in- 
formation passed on as it was in the original 

ICRAF stimulated the exchange and 
adoption of agroforestry technologies 
through village committees and meetings 
(barazas). The available data indicate that 
some of the villagers do not attend ICRAF 
meetings. They may not have time to do so. 
Women are absent more frequently than 
men. Women are taking care of children, 
milking cows, and doing other household 
chores, and such activities consume so much 
time that women cannot participate in the 
meetings (see Chapter 3). Furthermore, 
church meetings also constrain women; 
they must attend church on certain days. 
Funerals are similar: If you fail to attend, 
people say you do not think highly of the 

When lack of time is probed through 
observations, however, it appears that often 
people just sit at home instead of attending. 
This non-participation may indicate a lack 
of interest, and/or could be associated with 
the "agents" issue addressed earlier. A huge 
stimulant to attend ICRAF meetings, 
though, is the provision of "good food and 
drinks." 16 The data also hint at issues such 
as "people that attend such meetings with- 

hold the information for themselves." An- 
other is that announcements of meetings 
reach only a few and that only some people 
are invited. 

A group discussion in Isikhuyu showed 
that people perceive the whole concept of 
innovation as an external and superior 
practice. Thus, this knowledge necessarily 
has to be brought in by knowledgeable and 
unique persons. Some of the people used as 
agents and entry points into the community 
were previously working at the ICRAF 
demonstration plot in the village. That some 
of these agents had been watchmen at the 
demonstration site, however, contradicted 
the view that good knowledge comes from 

The fact that adoption is pervaded with 
ongoing social processes suggests that the 
success of these SFR technologies is de- 
pendent on the entire social framework 
within which it takes place. For instance, it 
was reported that attendance at training ses- 
sions is dependent on who makes the invita- 
tions and who else is likely to attend. Some 
of the characteristics found to be important 
include political alignment and perceived 
social status. 

A Typology of Users 

At this point we can elaborate in more detail 
what kind of typology of agroforestry users 
is relevant. The available data suggest four 
types, explained in this section. One sur- 
prise may be that these social typologies or 
portraits are distributed across the villages 
and are not distinguishable by access to 
resources such as land, labor, and capital. 
The quantitative analysis discussed ear- 
lier underlines that it is not easy to find con- 
sistent results linking adoption to particular 
assets or resources. Access to land is found 
to be important within the pilot villages, but 
wealth has no effect. In the non-pilot village 
analysis, however, almost the exact opposite 

16 The common meeting time became the late morning, as attendance would be highest (villagers no doubt 
manipulated this by not showing up at other times). ICRAF provided bread and sodas to participants. 


results are obtained. The typologies hinge 
much more on identities of people; that is, 
people's perspectives of how they see the 
future and how they identify and use new 

The " Seed Adopter." This social category 
refers to those who saw the opportunities 
offered by the seed market in the early 
phase of the project. The agroforestry- 
based SFR technologies require significant 
amounts of seed (crotalaria, for instance, 
is free seeded at a rate of 26,000 seeds per 
hectare). Thus, one of the driving forces to 
adopt SFR technologies was clearly the 
view that seeds are a source of cash. The 
relatively high prices of seeds at the time 
stimulated these types of "adopters" to grow 
seeds needed by ICRAF. Some of them 
managed to invest this money in other ac- 
tivities such buying dairy cattle. Most of 
these "seed adopters" lost interest in agro- 
forestry and dropped out of seed provision, 
however, as soon as seed prices dropped 
and the seeds were no longer collected by 
formal organizations. The issue of money 
in these cases overrode the soil fertility re- 
plenishment agenda of the SFR project. One 
old farmer said, "I have kept these seeds 
waiting for the time they will ask for seeds 
so that I can sell to them." 

In sum, it was not so much the soil im- 
provement effect that attracted farmers to 
plant such trees — despite the fact that the 
yields obtained from the same plot within 
the same period were better. As a result, 
when prices went down, a large number of 
people cut down their fallows. 

The "NGO-Networker. " This type of agro- 
forestry user stands for the individuals that 
through their early involvement with agro- 
forestry and ICRAF managed to maneuver 
themselves into strategic positions to gain 
access to resources distributed by NGOs and 
other projects or programs. Their involve- 
ment with agroforestry in their capacity of 
village elder or secretary of a community 
committee made them known to other 

agencies. The case material and other qual- 
itative research in the region (e.g., Mango 
2002) show that because of this, some of 
the "agents" got a dairy cow from a dairy 
development project, or a bicycle or other 
reward. In one of the cases in this project, 
after a man got a dairy cow, he clearly lost 
interest in agroforestry. As a village elder, he 
still keeps his agroforestry plots for demon- 
stration purposes, but his fields show that 
the technology is not used. Mango (2002) 
refers to a man who, because of his role as 
an adaptive research farmer for one of the 
NGOs, was nominated and later appointed 
assistant chief — a relatively well-paid job. 
After he took office, he completely neglected 
both agroforestry and adaptive research and, 
instead, married a second wife. 

The "Keeners. " This category includes 
those who perceive agroforestry as a useful 
way to replenish soil fertility. Many of the 
respondents who had used SFR technolo- 
gies said that they obtained progressively 
increased yields and appreciated the tech- 
nology. However, they typically failed to 
remember their prior yields. Many could 
remember only the previous season's yields 
and, more specifically, the portion of pro- 
duction that they managed to keep in the 
sack at the end of the season. Produce con- 
sumed while green in the field was difficult 
for them to estimate. 

These adopters are enthusiastic about 
agroforestry because it increases yields and 
reduces monetary costs for maintaining soil 
fertility. A reduction in the "hunger period," 
the medicinal value derived from some of the 
shrubs, and improved life styles resulting 
from raised incomes are clearly positively 
associated with the technology. A second 
important element in their explanation is 
that agroforestry- based technologies can be 
easily combined with other technologies, 
such as the use of farmyard manure, cow 
dung, and compost. 

Dis- and Non-Adopters. Grouped together, 
dis- and non-adopters are the farmers who 


never tried or have stopped using agroforestry 
because of labor requirements, age, short- 
age of land, and/or the small size of fields. 
A variety of people fit this typology, which 
analytically and practically makes this the 
most difficult group to identify. Generally, 
though, the group can be subdivided into: 

1. Those who do not adopt because of 
age. In the qualitative sample, it was 
mostly widows who fit this profile. 

2. Those who have difficulty rallying and 
controlling family labor, and thus can- 
not meet the extra labor input required. 
The case studies show that this profile 

Families in which spouses are labor 

Families affected by HIV/AIDS across 
the social unit 

Families that have difficulty in con- 
trolling family labor because of 
intergenerational differences 

The latter two deserve more detailed com- 

performing casual labor on someone else's 
farm, rather than his parents'. Young women 
mostly work together with their parents. 

There is another generational difference 
to explore. Youth generally perceive farm- 
ing differently from their parents. The youth 
who pursue farming plant high-value crops 
such as vegetables or chewing -type sugar- 
cane to sell. This group prefers using com- 
mercial fertilizer in farming rather than 
improved fallows and tithonia. In contrast, 
their parents produce crops for subsistence 
first and what is left (if any) goes to market. 

Another issue is that, despite the small 
land size, social expectations for young 
people affect their attitude toward farmwork 
and SFR technologies. The society expects 
them to migrate to the urban areas, work, 
and send money home. Therefore any youth 
who cannot do this may feel like a failure 
and therefore find it difficult to remain in 
the rural area. To accomplish this social ex- 
pectation, most of the youth have become 
migrants and thus are not present in the 
villages where farming takes place. 

HIV/ 'AIDS- Affected Families. These fami- 
lies lose labor because of AIDS, which, in 
the majority of cases, affects the most able 
people, fathers and mothers. Financial re- 
sources dwindle to pay medical bills and are 
not allocated to hiring labor for agricultural 
purposes. HIV/ AIDS -affected families are 
among the most fragile in the communities, 
for whom labor-intensive agricultural work 
is most difficult. 

Generational Issues. Many of the male 
youths, out of school but not yet married — 
with no direct responsibility to feed some- 
one else — have no motivation to work in 
the fields. They are focused on finding white- 
collar jobs in town — they "tarmac" as it is 
called locally. If the job has to be in the agri- 
cultural sector, then for them it must pay: 
their labor culture is not tuned to working 
collectively on their parents' land for a 
pooled yield at the end of the season. Their 
parents, however, will not pay for their labor. 
Therefore it is common to find young men 

Fertilizer and Cattle Manure 

Table 6.11 shows the relative importance of 
the SFR technologies introduced by ICRAF 
and others in the region. Animal manure 
and fertilizers are used more frequently than 
fallows and biomass transfer technologies. 
The use of manure and fertilizer has in- 
creased steadily over the last few years, 
partly due to some users' strong belief in 
fertilizers. Wilbert from Sarika village, for 
example, commented that once "you have 
started using fertilizers, it becomes very dif- 
ficult to do without it, because it is not pos- 
sible to realize good yields without applying 
it." When he plants with DAP, the growth 
rate of plants is high and, according to him, 
better than results using tithonia. 

Others, like Joseph from Ishikhuyu, said 
that he has never used any inorganic fertil- 
izer because he believes his land is still fer- 
tile and maize yields are fairly good. But, on 
the other hand, he says, "I don't have enough 
money to purchase any inorganic fertilizers." 


Table 6.11 Use rates of soil fertility management options over time in 
non-pilot project areas 

SFR options 






Animal manure 












Improved fallows 






Biomass transfer 






The case material suggests that fertilizers 
are mostly used for commercial crops like 
sukumawiki (kale) and tomatoes, rather than 
food crops (maize, beans, millet/sorghum). 

Analysis of the available quantitative 
and qualitative data shows that fertilizer and 
animal manure use are positively linked to 
wealth, both within and outside the pilot vil- 
lages. Fertilizer is also strongly linked to 
education levels, perhaps directly and indi- 
rectly through the implication of having had 
off-farm employment. All practices are 
positively linked to farm size to some de- 
gree in at least one regression. Fertilizer 
use is less common among female-headed 
households than male-headed households. 
Synthesizing these results suggests that 
improved fallows and biomass transfer may 
be reaching a wider clientele, including less 
advantaged groups, than fertilizer or animal 
manure options. This cannot be said of com- 
post, however, which appears to be quite 
common among women farmers and less 
educated farmers. An analysis of the pilot 
villages found that the agroforestry tech- 
nologies were being adopted or tested by 
44 percent of those households that were 
not using other soil fertility methods. In 
other words, they seem to be extending the 
range of options rather well. The compa- 
rable figure in the non-pilot villages is lower, 
but still encouraging, at 30 percent. 

The relevance of alternatives such as bio- 
mass transfer and improved fallows as soil 
fertility management options is that both 
fertilizer and animal manure application is 
constrained by a number of factors. One 
factor as already indicated by Joseph is the 
price of fertilizer, which has increased pro- 
gressively over the last 20 years or so. The 

majority of people interviewed commented 
that cost prevents them from using fertilizers. 
In addition, fertilizer prices increase faster 
than produce prices farmers can charge. 
This squeeze on agriculture (also labeled as 
"the treadmill") is not likely to decrease in 
the near future. It is therefore unlikely that, 
in poor areas such as western Kenya, fer- 
tilizer will remain an attractive option. Fur- 
thermore, in approximately half the cases, 
farmers insisted that "fertilizer spoils the 
soil." They sincerely question the effective- 
ness of fertilizer as a way to reproduce soil 
fertility. However, such an opinion is con- 
tested by other farmers, who claim that fer- 
tilizer is a perfect technology, though unfor- 
tunately not affordable for them. Issues such 
as these suggest that fertilizer use cannot 
be explained without reference to poverty, 
rumors, and misunderstandings about nutri- 
ents, flows, and soil-plant interactions. 

Similar factors can be mentioned when 
discussing the application of animal or 
boma manure. Animal manure is a tradi- 
tional favorite as far as soil fertility replen- 
ishment is concerned. The Luo, in particular, 
were pastoralists and traditionally valued 
cattle. Cattle rearing and arable farming 
were linked by cattle providing nutrients for 
arable agriculture. Over the years, however, 
the number of cattle has dropped substan- 
tially owing to lack of grazing area. Tech- 
nologies developed over the last three 
decades and improvements in the field of 
zero grazing, though, now provide work- 
able alternatives for scarce land. Feed and 
fodder systems — such as cut-and-carry of 
napier grass and the growing of fodder trees 
— reduce the demand for grazing land. This 
may in the end help fulfill the dream of Luo 


and Luhya men and women to own dairy 
cattle to restore the connection between 
livestock and arable farming. Manure now- 
adays is much favored, as it does not require 
cash outlay as long as one owns cattle. Only 
occasionally do villagers purchase manure. 
Salome's story shows that there is more to 
add to the analysis of manure and inorganic 
fertilizer use and that one has to include 
property rights as well. Salome, who lives 
in Arude in Luoland, mentioned that she 
used to add boma manure to her land where 
she cultivates continuously. She did this 
until 2000, when she fell sick. Thus she says 
she is likely to realize reduced yields this 
year. She has never used any commercial 
fertilizers because (1) they are expensive 
and she lacks funds to purchase them and 
(2) she now has the alternative of using land 
owned by the extended family. When a field 
is no longer productive, she can open a new 
field (which is not easy for her at this time 
as she only occasionally gets help to clear 
the bush and remove the stumps of trees). 
She also commented that when she was 
young, boma manure was never applied on 
the fields, as these were then very fertile. 
Instead, cow dung, she said, was used in the 
past as fuel in the cowshed, to cook long- 
cooking meals such as blood or nyoyo 
(maize and beans), and to provide warmth 
to people and animals out in the open. 

The comparison of fertilizers and ma- 
nure with improved fallows and biomass 
transfer is made not only with reference to 
prices. Issues of labor (particularly in the 
case of biomass transfer) and land size 
(very relevant for improved fallow) play a 
role as well. Fertilizer certainly has the 
advantage of being less labor-intensive than 
biomass transfer. These issues have been 
dealt with in previous sections of this chap- 
ter but should be taken into account when 
all the soil fertility replenishment practices 
are examined and considered. 


It is significant that, once adopted, SFR 
technologies are adapted or redesigned to 

fit a farmer's particular situation. Concern- 
ing improved fallows, farmers claim that 
C. grahamiana trees are more difficult to 
handle compared to Tephrosia vogelii or 
any other Tephrosia species, because C. gra- 
hamiana can attract large caterpillars. How- 
ever, farmers have discovered that inter- 
planting the two tree species reduces or 
eliminates caterpillar infestations. Therefore 
many farmers interplant them on the same 

Farmers also noted that Tephrosia roots 
tend to keep away moles; this is another 
reason for interest in planting more than 
one species. Farmers also played a role in 
the eventual definition of the "best-bet" fal- 
lowing systems. Initially researchers had 
experimented with Sesbania sesban, which 
required a nursery stage before planting 
seedlings on a fallow field during the onset 
of the rains. Farmers preferred species that 
could be sown directly, saving considerable 
labor time. They further helped develop a 
system whereby the tree seed is sown into 
an existing maize crop. This reduces the 
amount of time land is taken out of produc- 
tion and also cuts the labor needed to weed 
a separate tree plot. 

With respect to biomass transfer, there 
has been little innovation in the species 
used, and tithonia still dominates. What has 
changed as a result of farmer innovation is 
the range of crops on which biomass trans- 
fer is used. While researchers had conducted 
all their research on maize and beans, farm- 
ers tried tithonia on kales, tomatoes, and 
French beans. In the villages, a significant 
and growing number of farmers have 
switched to using tithonia on higher value 
crops. This trend has also influenced dis- 
semination messages, which now emphasize 
the greater returns to biomass transfer on 
high-value crops. Farmers have also been 
experimenting with tithonia leaves. They 
sometimes add them to compost heaps to 
produce higher quality material. They have 
also made liquid nutrient concoctions, which 
they find easier to apply to the soil. Last, 
some have been testing the usefulness of 
tithonia as a pesticide. 


Soil Fertility Replenishment and Rural 
Peoples' Livelihood in Western Kenya 

This chapter provides insight into the complex relationship between two soil fertility 
management practices (improved fallows and biomass transfer) and livelihood trans- 
formations and improvements. While the previous chapters and notably Chapter 6 have 
accounted for the processes that explain adoption or dis-adoption, here we aim to address 
whether such soil fertility management interventions have had a positive impact on rural peo- 
ples' lives. In the discourse of the sustainable livelihood (SL) framework, this chapter deals 
with the outcome indicators of livelihood improvement-related interventions and activities. 

One of the advantages of the more ethnographic approach of this research is that it allows 
us to introduce the voices of participants in the interventions. The first section brings this to 
the fore, with the rest of the chapter referring to both qualitative and quantitative data analy- 
sis. The second section considers the anticipated relationship between soil fertility and crop 
productivity, especially for maize. The third section examines the quantitative relationships 
between the SFR technologies and changes in a number of household-level welfare indicators. 
The indicators explored are related to household assets, nonfood expenditures, and food con- 
sumption and nutrition. The patterns of changes in these variables are described first, and these 
descriptions are followed by econometric analyses to explain these changes. The following 
section analyzes how SFR impacts are conditioned by many individual and household factors 
using the case study evidence; in other words, who has benefited from the SFR technol- 
ogy interventions and what are the constraints preventing impact on a wider population? The 
penultimate section attempts to position the impacts from SFR technologies within the broader 
context of poverty alleviation in western Kenya. The final section summarizes the key findings 
of the chapter. 

Voices and Realities 

The major incentives to use these technologies appear to be income from the sale of seed, in- 
crease in yields, reduction in the "hunger period," the medicinal value derived from some of 
the shrubs, and improved lifestyles due to raised incomes. 

Farmers state that the various SFR technologies have increased their yields, raised house- 
hold incomes, and improved food security or their ability to mitigate crises. But, as we will 
see from the two case study accounts later, actual impact also depends on the circumstances 
in which these SFR technologies are applied. Both Gilbert and Asselo are retired, older men 
faced with challenging situations, including eking out a living from fairly unproductive farm 
activities. Their case history accounts show that biomass transfer and fallow crops have im- 
proved their farm yields, especially for food crops. Gilbert reports an output of four bags of 



Box 7.1 Increased Farm Yields 

Gilbert retired and returned to his home village in 1993 after about 50 years in Nairobi. 
Since then he has been working on the farm together with his wife, Hellena. They have 
four grown children, all married. The couple registered themselves as ICRAF farmers 
about three years ago. 

Gilbert owns 1.3 hectares of land and one cow. Most of the land is on a slope lead- 
ing down to a stream. He says this land has not been productive and requires soil fertility 
improvement. Although he had heard of ICRAF's ideas earlier, he did not adopt until 
the year 2000, when he decided to plant C. grahamiana and T. vogelli. He obtained 
seeds from ICRAF. The major incentive for him to plant more C. grahamiana is that 
ICRAF buys seeds at a good price. However, so far, he has not been able to sell seeds to 
ICRAF because his plants are not mature yet. Some people have already stopped plant- 
ing C. grahamiana because it has been some time since ICRAF purchased seed from 
farmers. In addition, C. grahamiana can attract large caterpillars and is therefore not 
popular with many people. 

Gilbert is convinced that these fallow crops add "manure" to the soil because al- 
though he has not planted food crops on this piece of land, he is able to tell from the type 
of weeds that grow now. They look healthy. He intends to clear and dig it when he gets 
capital to hire labor. 

He has also planted tithonia on fanyajuu terraces. He decided to use tithonia because 
he saw a neighbor get good yields from a small piece of land. And when he used titho- 
nia as green manure, he harvested four sacks of maize from two plots where he used to 
get about two. He now intends to plant more tithonia because he has realized that it gives 
good yields. However, because of their advanced age, Gilbert and Hellena may not prac- 
tice these technologies for long. Already, Hellena is not keen on the SFR technologies 
taught by ICRAF because she feels that the work is too hard. 

Gilbert plants indigenous varieties of maize and beans because he finds hybrids to 
be unaffordable. He used DAP once in 1994, but says that fertilizers make soils un- 
productive and salty in the long run. He has also used farmyard manure, but most of his 
cows were stolen; the remaining one cannot provide enough manure for his farm. 

On the other hand, Hellena, Gilbert's wife, sells fermented finger millet {thowi) at 
Yala market twice a week. She spends her profits on foodstuffs and occasionally hires 
labor to work on the farm. Because Hellena cannot carry the load to and from the mar- 
ket, Gilbert assists by transporting the thowi on his bicycle. When he gets committed 
elsewhere, they organize with a nephew or any other bicycle transport to take it to the 
market. Hellena explained that unlike farmwork, she could still carry on with the thowi 
business because "I can send someone like my husband or my daughter-in-law or any 
other relative to buy for me the dried finger millet from the market. I would be able to 
do other things and some of my customers would come to buy from me at home. This 
business helped me a lot during that period. Even now that I can go to the farm a bit, it 
is still good because the farm yields get finished before the next harvest and we can use 
this to buy food. Actually, when I was sick, we used the little savings I had from this 
business to seek treatment. This has reduced my capital investment in the business, and 
currently I cannot buy ten gorogoros of finger millet as before. But I hope God will help 
me get more money to reach where I was before." 


Sometimes Hellena and Gilbert receive financial assistance from their younger son, 
who is employed at a fairly stable job in town. Their eldest son lives with them at home 
and, they say, is not responsible. He spends much of his time in politics, and his wife is 
now also dependent on Hellena and Gilbert. In fact, Hellena complained that this son 
does not even send his children to school, claiming that he has no money. Hellena in- 
tended to pay their school fees so that they could go back to school. Because they some- 
times have nothing to eat in their house, Hellena shares what they have with them. 

According to Hellena, a poor person is one who cannot carry out his or her farmwork 
effectively because of a lack of income and new ideas. Such a person has no food to eat 
and cannot send his children to school: "Look at my grandchildren here. They cannot go 
to school because their parents cannot afford it. Now I am trying to work hard to get for 
them some money to send them back to school." According to Gilbert, people in his 
village have different economic abilities. And, in his assessment, he is neither rich nor 
poor because he can afford to work on his own farm and harvest something for his food. 
Nevertheless, he feels that it is important to have another source of income besides farm- 
ing, especially when seasonal rains change, as has been the case in the recent past. 

Gilbert and Hellena concur that a rich person is one who has money, plants his own 
food and has surplus, owns livestock, and has new ideas that can be implemented suc- 
cessfully. They say that this is only possible with adequate money. Hellena, however, 
stressed that money alone does not equal riches, especially when it is not used well. 

Source: Sarika Village, Siaya District. 

maize after using SFR, up from just two 
bags produced previously (Box 7.1). Asselo 
(Box 7.2) claims similar output. However, 
even when yields double, the net effect may 
not be substantial. This is because for most 
of these households, needs are many and 
acreage is limited. In other words, although 
the SFR technologies do improve produc- 
tivity, they cannot on their own dramatically 
reduce poverty. 

For instance, Asselo sold more than half 
of his maize harvest, but the earnings trans- 
late into just 2,400 KShs. This cash went 
immediately into basic needs such as buy- 
ing sugar for tea and paying school fees. 
Furthermore, we are not convinced that the 
two bags set aside for home consumption 
were sufficient for his food needs. Low pro- 
ductivity and low market value of cereal 
output bring about food scarcity; this is 
exacerbated by the need to sell a sizable 
portion of the maize harvest. So even in sit- 

uations in which improved yields may have 
contributed to stabilized supply, improved 
food security may not be observed. 

Nevertheless, adoption of SFR tech- 
nologies is also contributing to training 
and the general development of the human 
capital base. Both Asselo and Gilbert have 
been able, through these technologies, to 
start new lives after retirement from formal, 
off-farm employment. It is possible that 
each of these men would have been quite 
disoriented if they were to return to the vil- 
lage to depend entirely on less productive 
farm activities. Both men have financial re- 
sponsibilities that they seem to meet from 
their farm incomes, however slight. 

The case studies also suggest that the 
SFR technologies adopted have given some 
members of the community an amount of 
social capital, especially in terms of their 
being seen as successful farmers and people 
who attract visitors from "far away." Indeed, 


Box 7.2 Mitigating Vulnerability 

"My name is Asselo and I am 52 years old. I schooled up to Form Two before I was 
forced to drop out of school in 1972 for lack of fees. Between 1972 andl974, 1 assisted 
my mother on the farm, planting maize, bananas, and coffee. In 1975, a friend invited 
me for a job in Eldoret as an apprentice up to 1982, when I quit the job because I was 
sickly. I stayed at home for a year seeking medical treatment. By the end of 1983, 1 had 
become so broke and idle that I had to get something to do. A great friend of mine whom 
I once assisted to find a job gave me a soft loan of KShs 20,000 with which I opened a 
shop at Soy in Lugari. The business was good and I earned about KShs 3,200 profit per 
month. I used this money to educate my three stepbrothers and my children in primary 
school. I also used to hire land and plant maize, which I then sold at a higher price. I 
used these earnings to buy land at the settlement scheme. 

"In 1996, the land on which my shop stood was demarcated and allocated to squat- 
ters, and I was forced out. In November of the same year, my mother passed away and 
this compelled me to return to the village to take care of the home. I had no option be- 
cause I am the last-born, and I had stayed away from home for 22 years, and age was 
catching up with me. 

"In 1997, 1 started farming maize, beans, cassava, and sweet potatoes. These were the 
easiest crops and everyone else was growing them. I harvested five bags of maize and 
two bags of beans. I never used any fertilizer because I had no money, and the majority 
of the people were not using any fertilizer to plant maize. I saved two bags for home con- 
sumption and sold three bags at KShs 800 each. This money went into buying sugar, 
books, and paying school fees for my last-born child in secondary school. In 1998, I 
planted maize, beans, onions, and sukuma wiki. 

"In 1999, 1 got grahamiana from ICRAF when they came to see my friend, Fannuel. 
I planted the trees on a small piece in my shamba and immediately they reduced the 
Striga weed. After removing these trees, I planted maize, and there was a difference as 
compared to the other years. I harvested five bags of maize from half an acre of land 
when previously I used to harvest only half a bag of maize. I have also learnt from my 
neighbors how to make compost manure. I find it easy and cheap to prepare because I 
don't need to hire any labor or buy inputs. 

"This is different from before, when at the settlement scheme I used DAP for plant- 
ing and urea for top dressing and harvested 35 bags of maize per acre. That time I had 
money from business. I have never tried this at home because I have no money. I am now 
a very poor man." 

Source: Isikhuyu Village, Vihiga District. 

some of these visits have been so eventful 
that several families have named their chil- 
dren after these personalities. On the other 
hand, the decision to adopt or not to adopt 
SFR technologies as a livelihood strategy 
has brought about jealousies and disagree- 
ments, some of them at the level of the 

family unit. In the case described earlier of 
the husband and wife who now pursue dif- 
ferent farming practices, the use of these 
SFR technologies has changed the hus- 
band's status. This particular farmer is now 
described by others as someone who has 
adopted the SFR technologies successfully. 


On the other hand, the potential of some 
of the SFR technologies is realized on only 
a few farms, largely because most of the tar- 
get farmers are resource poor. For instance, 
although Gilbert is already aware of some 
of the potential benefits of fallow crops, he 
was yet to cultivate a piece of land that had 
been under fallow crops owing to lack of 
labor to clear the field. He himself is an old 
man, and his equally old wife is unable to 
assist because of an old injury. For the same 
reasons, Gilbert is not so interested in bio- 
mass transfer. 

It is also evident from Asselo's account 
that people tend to take up these SFR tech- 
nologies — in fact, farming in general — after 
all else has failed. By this time, they may be 
too old or too poor to invest what it takes 
to realize good yields. So the agroforestry 
technologies appear to be providing some 
scope for development among disadvan- 
taged households. However, the success of 
these SFR technologies depends on having 
in place some minimum provisions. Indeed, 
findings show that higher yields and raised 
incomes have not always translated into 
improved economic well-being, at least as is 
commonly understood (e.g., increased food 
consumption). In some instances, additional 
production could not be properly stored or 
was sold at very low prices under distress. 
In a few instances, increased incomes have 
resulted in a man taking a second wife. This 
might lead to enhanced welfare of the man, 
but not necessarily that of the first wife or 
other household members. 17 Hence, bene- 
fits vary across gender and accrue to differ- 

ent members of the household. Some house- 
holds have succeeded nevertheless. 

SFR, Production, 
and Productivity 

In this section, we present the results of 
many analyses linking the use of SFR and 
crop yields. This section is distinct from the 
econometric analyses in the following sec- 
tions because unlike other indicators, no 
baseline yield estimates were collected from 
farmers' fields. 18 We include data from our 
surveys of farmers, the case study reports, 
and researcher-designed/farmer-managed 
trials. Table 7.1 presents a summary of the 
performance of several soil fertility replen- 
ishment practices on maize yields, based on 
farmer recall in the non-pilot villages. The 
number of farmers using the practices is 
higher than the number used to calculate 
yield effects because of certain missing or 
unclear data. The SFR yields are compared 
to a control case of maize production with 
no soil nutrient inputs. 

It can be seen that fertilizer, improved 
fallows, and biomass transfer all led to 
positive yield changes in most cases, with 
fertilizer being the most likely (93 percent) 
to lead to a positive change, but improved 
fallows and biomass transfer close behind. 
All three SFR practices were reported to 
have significant effects on yields, as reported 
by the percentage increase in median and 
mean yield. 19 For example, median yield in- 
creases from biomass transfer and improved 
fallows were equal to 167 percent over a 

17 These findings support other studies that have found social networks to be very important to the diffusion of 
other types of innovations in the region. For example, informal women's groups were found to have facilitated 
adoption of birth control where cultural values and beliefs discouraged adoption, and where other programs had 
failed (Rodgers et al. 2001; Behrman, Kohler, and Cotts Watkins 2002). 

18 In practice, establishing a baseline would have been tricky, because farmers would select a portion of a large 
maize plot on which to try a fallow. Because there is within-plot heterogeneity, it would have required yield mea- 
surements at very small scales. 

19 We report the percentage increase and not the absolute increase because plot areas appeared to be rounded up- 
ward in many cases, implying that absolute yields would be biased downward. 


Table 7.1 Soil fertility practices and maize yield impacts 

Improved fallow 

Biomass transfer 


Number of cases 




Compared to no inputs: 

Percentage with non-positive effect 
Mean increase in yield (percent) 
Median increase in yield (percent) 
Median size of area (acres) 










no-input maize cropping system. The agro- 
forestry practices compare favorably with 
fertilizer because fertilizer amounts are quite 
low in practice, whereas farmers are able 
to generate significant amounts of nitrogen 
from the agroforestry systems on the rela- 
tively small plots on which they were ap- 
plied. The last row of Table 7.1 shows the 
median size of field on which the practices 
are applied. While the typical farmer who 
uses fertilizer applies it to one acre, the area 
under the agroforestry systems remains 
considerably lower, at 0.25 acres (recall that 
the more accurately measured fallow fields 
in the pilot villages indicate an average of 
about 0.11 acres). 

We tried to calculate a financial analysis 
of the data reported in Table 7.1, but this 
proved unsuccessful. First, the data on plot 
area are imprecise and absolute differences 
in yields therefore not reliable. Further, be- 
cause labor data were not collected from this 
single-visit survey, precise plot areas are 
required to apply average labor figures from 
secondary sources. Finally, for biomass 
transfer systems, farmers were unable to 
provide accurate information on the amount 
of biomass collected and applied. 20 Instead, 
farmers in the non-pilot villages were asked 
to describe how their maize yields have 
changed between 1997 and 2001. These 
changes were compared against the use of 
SFR practices and wealth level of house- 
holds. These measures of yield change are 

not as precisely attributable to SFR be- 
cause we asked for general impressions, but 
given the generally pessimistic attitude of 
many farmers in western Kenya, the associ- 
ation between SFR practices and improved 
yields is potentially important to spurring 
development processes. In terms of per- 
ceived changes in maize yield, 36.8 percent 
were thought to have remained unchanged, 
28.5 percent were lower, and 34.7 percent 

Cross tabulations were run contrasting 
the use of different SFR practices and per- 
ceived maize change over the 1997-2001 
period. Fertilizer is most strongly associated 
with perceived positive maize productivity 
change. Households who are frequent users 
of fertilizer rarely report declining maize 
yields (8.5 percent), whereas 35.2 percent of 
households who do not use fertilizer report 
decreasing yields. There are similar, posi- 
tive patterns with biomass transfer, improved 
fallows, and animal manure, but they are 
not statistically significant. 

Perceived maize yield change is not 
highly linked to wealth group. The enumer- 
ator evaluation measure of wealth/poverty 
was, however, very significant: 47 percent of 
the wealthy stated that their yields increased 
as opposed to only 23 percent of the poor. 

On a related issue, farmers were asked 
to provide data on the area under maize in 
2001 and 1997. One hypothesis is that if 
SFR can improve yields, it may catalyze 

20 Thus, we are collecting more detailed information from pilot village farmers for the 2003 long-rain season for 
subsequent analysis. 


shifts into other higher value crops. The 
data show that 67 percent of farms did not 
change area under maize, 9 percent had de- 
creased, and 24 percent had increased area. 
Analysis of variance (ANOVA) and correla- 
tion tests did not find any relationship be- 
tween change in maize area and the use of 
biomass transfer, improved fallows, or com- 
bined use of SFR practices. 

Within the pilot villages, two more 
controlled analyses were made. First, we 
present findings from farmer-managed trials 
of improved fallows and biomass transfer 
within the pilot villages. The improved fal- 
low trial involved about 70 farmers, and 
yields from control and treatment plots were 
carefully measured by technicians for four 
consecutive seasons. The control was the 
planting of maize with no nutrient inputs in 
every season. The improved fallow trials 
involved one or two seasons under trees 
with two or three seasons of maize (so that 
the opportunity cost of land is included in 
the calculation). The seasonal per acre net 
gain to tephrosia fallows was $22.33 and 
for crotalaria it was $19.96 (again, com- 
pared to the no-input case). The same set 
of trials also assessed the returns to labor 
from fallowing systems that were found to 
be around $2.17 per day, 33 percent higher 
than from no-input, continuous maize pro- 
duction. Returns to biomass transfer on 
maize in trials fared poorly owing to high 
labor costs against relatively low-value re- 
turn from maize, but amounts of biomass 
in the treatments were substantially above 
those commonly applied by farmers. 

These returns are not large, but they are 
important to very poor households. To find 
more profitable opportunities, farmers have 
directed soil nutrient inputs to higher value 
crops, rather than maize. Farmer-managed 
biomass transfer trials with kale and toma- 
toes have shown that similar increases in 

yields are obtained on these crops. Because 
they fetch much higher prices, returns to 
land are much higher than on maize. For ex- 
ample, returns to biomass transfer on veg- 
etable production were high, with returns to 
land reaching as high as eight times those 
with no nutrient inputs. 21 

A second analysis is a production func- 
tion estimation involving many of the same 
households as covered in the household 
welfare impact analysis presented later, but 
for the 2003 long rains production season, 
the year after the main study was com- 
pleted. The data on yields and inputs are 
from farmer estimates; although many of 
the data are reasonably collected, there are 
problems with the biomass transfer and im- 
proved fallow variables. There were only 10 
positive biomass transfer values among 
150 plots. For the improved fallow variable, 
about 30 positive cases were reported. Yet, 
from our annual monitoring survey (see 
Chapter 6), far more households had in- 
vested in fallows, so it is unclear whether all 
are captured in this new data set. Moreover, 
the average size of plot in this estimation is 
0.14 hectare, which is the same as the aver- 
age size of plot that was recently under an 
improved fallow. However, our other data 
set is quite clear that the average size fallow 
was only 0.04 hectare (see Chapter 6). Thus, 
it is uncertain as to whether the measured 
yields actually correspond uniquely to the 
effect of the fallow. For all these reasons, 
the reliability of these results is question- 
able. For biomass transfer, we found a pos- 
itive and significant result (p = .04). The 
coefficient value was extraordinarily high, 
beyond reasonable effects based on nutrient 
content of the biomass, and thus is difficult 
to explain. For improved fallows, the coef- 
ficient was not statistically significant. As 
noted earlier, it is unclear whether the fal- 
low plot and maize yields were properly 

21 There are many varied treatments using combinations of organic and inorganic nutrients with few observations, 
so it is dangerous to report means. However, returns to kales with tithonia range between $600 and $ 1 ,000 per 
hectare per season. 


matched up through the survey. It is obvious 
that trying to assess yield effects of SFR is 
fraught with many challenges. 22 

Econometric Models 
to Evaluate Household 
Impacts of SFR 

In the following three sections, econometric 
models are used to assess the effect of bio- 
mass transfer and improved fallow systems 
on changes in asset values, changes in non- 
food expenditure, and changes in food and 
nutrition indicators. All of these variables 
were tested on our pilot village sample. In 
the non-pilot villages we were able to in- 
vestigate only changes in assets because of 
lack of a baseline for other variables. How- 
ever, this analysis is not reported, owing to 
lack of sufficient instrumental variables 
taken prior to the period of adoption of agro- 
forestry. The rest of this section therefore 
relates solely to the pilot villages and that 
sample of households (n = 103). 

The testing of the effect of the use of 
improved fallows and biomass transfer is 
not straightforward because they are also 
endogenous variables. Hence, two-stage 
methods must be employed in which adop- 
tion of agroforestry is explained in the first 
stage and the predicted values used in the 
second-stage impact regression. One require- 
ment for this analysis is the identification of 
variables that may affect adoption intensity 
but not impact. This is not easy to do from 
a theoretical aspect, because adoption and 
impact on assets are very closely related. 
The variables we selected as instruments to 
explain adoption but not impact relate to 
household perceptions of the importance of 
agroclimatic shocks for the village (the 
risks of drought, hail, and pests/diseases 
perceived prior to the period of study), the 
father's farm size, and the jobs and social 
positions held by their fathers. We also in- 

cluded whether either of the adult members 
of the household had previously held a job 
in the formal sector. Some of these might 
also affect impact, but we feel that those 
would be much more muted than the effects 
on adoption because they relate mainly to 
incentives to adopt or exposure to new 
ideas, but have little to do with direct agri- 
cultural management, which is important 
for adoption to translate into impact. 

It is expected that where drought is per- 
ceived as a common occurrence, there is 
less interest in fertilizer (which is known to 
perform poorly during droughts) and thus 
more interest in fallows or biomass transfer. 
Occurrence of hail, pests, and diseases may 
have the opposite effect on interest in agro- 
forestry. A father's farm size is expected to 
influence the farming practices learned as a 
child. Where farm size is smaller, it is ex- 
pected that intensification practices, includ- 
ing fertilizer use, were advanced and picked 
up by the current household. On the other 
hand, larger farms would have been more 
likely to practice fallowing and less likely 
to use fertilizer, leading to reduced under- 
standing of fertilizer for the current house- 
hold. It is expected that where the father had 
a good non-farm job or acquired a position 
of influence, he may have been more likely 
connected with extension agents and the 
private sector and therefore been farming 
with fertilizers. Much like the father's expe- 
rience, asking about previous formal-sector 
employment of the head or spouse attempts 
to capture his or her access to knowledge of 
modern farming techniques such as fertiliz- 
ers. Although these variables are hypothe- 
sized to influence choice of soil fertility 
management practice, we do not expect any 
of them to be associated with actual impacts. 

A second issue to resolve is the mea- 
surement of adoption of the agroforestry 
systems. Recall that in the previous chapter, 
adoption was described by noting different 

22 Fertilizer quantity had a positive and significant effect on yield, but the effect of manure quantity was found to 
be insignificant. 


dynamic patterns, for instance, dis-adoption 
and continuous use. This was very relevant 
in trying to capture quite distinct dynamic 
behavior. Here we are less concerned about 
teasing out the different motivations behind 
the use of agroforestry but rather interested 
in how different degrees of use may affect 
asset holdings and other welfare indicators. 
This is better measured by a continuous 
variable that can capture intensity of adop- 
tion over space and time. For improved 
fallows, the sizes were verified by enumer- 
ators in the pilot villages. The sizes of plots 
on which biomass transfer was used were 
not measured. 23 Hence, for the pilot vil- 
lages, our "intensity" variables are the sum 
of total area under improved fallow and the 
sum of the number of seasons for which 
biomass transfer was used. Both of these 
include only those seasons relevant to the 
study period (six seasons). 

All these intensity variables are continu- 
ous variables, but they exhibit a non-normal 
distribution in that there are large concen- 
trations at the value of zero — which reflects 
all the households who never used the tech- 
nology. We therefore have two approaches 
we can use in the two-stage procedure. The 
first is to proceed and run both first- and 
second-stage regressions using ordinary 
least squares (OLS). This leads to biased co- 
efficient estimates in the first stage, owing 
to the nature of the dependent variable, but 
the standard errors for the predicted values 
in the second-stage regression are unbiased. 
This is computationally easiest, as it is 
simply a two-stage least-squares procedure. 
The second is to run tobit models in the first 
stage, which gives unbiased estimates in the 
first-stage regression. The predicted values 
from the tobit model can then be used in 
the second stage. However, in this case, the 
standard errors are biased and a technique 
such as bootstrapping needs to be employed 

to be able to give reasonable estimates of 
unbiased standard errors for the coefficients. 
Both methods are used and second-stage 
results from both are discussed. For im- 
proved fallow area, the correlation between 
the actual value and the predicted value is 
0.49 from the tobit model and 0.51 from 
the OLS model. The fit is not as close for 
biomass transfer, but is reasonable at 0.36 
from the tobit model and 0.38 from the OLS 

In the two-stage least-squares approach, 
three included variables (perception of hail 
problem, father's job, and father's social 
position in the community) were significant 
to the total area under improved fallows 
(see Appendix A for a full presentation of 
first-stage results, including statistical tests). 
However, none of the first- stage regression 
variables were significant in the number of 
seasons using biomass transfer. For im- 
proved fallows, the first-stage OLS regres- 
sions for fallow area were significant in 
terms of an F -test (p = .02) and the tobit re- 
gression chi-square value was fairly strong 
(p = .12). However, the regressions for bio- 
mass transfer were not highly significant. 
We ran a series of tests to check for the ef- 
fectiveness of the instrumental variables. For 
these, mixed results were obtained. Durbin- 
Hausman-Wu tests on the residuals re- 
vealed some remaining concerns on the 
correlation between improved fallow area 
and the residuals for the expenditure and 
iron nutrient regressions. There were no 
problems for the other impact indicators. 
Hausman tests showed that the instruments 
were acceptable in all cases except for the 
improved fallow variable in the expen- 
diture model. Last, chi-square tests for over- 
identification using a regression of residuals 
on all explanatory variables and instruments 
could not be rejected in four of the six im- 
pact regressions (overidentification could be 

23 The reason that they were not measured by enumerators is because they could not be observed in the field (the 
biomass is incorporated into the soil and not visible), unlike most fallows that were in the field during the time 
of annual monitoring. 


rejected only in the expenditure regressions). 
So, to conclude, the instruments selected for 
the most part seem to do reasonably well in 
terms of correlating with the actual value and 
cutting the correlation to the error terms. 
There are remaining concerns with the set 
of instruments, which is not surprising, given 
the difficulty in distinguishing between the 
close concepts of adoption and impact and 
perhaps also because of the small number of 
observations. There are particular concerns 
as to the usefulness of the predicted im- 
proved fallow variable in the expenditure 

Three models are run for each indicator. 
A two-stage least squares model is run with 
household variables such as age, gender, ed- 
ucation, and ethnicity of household head 
used as explanatory variables. 24 The house- 
hold variables reflect pre-adoption values 
(in reality, few variables changed over time 
anyway). A second model is a difference 
equation and tests for the link between agro- 
forestry use and the welfare indicators, fac- 
toring out household structural factors. The 
third model is a two-stage approach in which 
tobit models are used in the first stage to 
generate predicted values of the intensity of 
agroforestry use. In the second stage, boot- 
strapping techniques are used to improve 
the estimates of the standard errors of the 
coefficients. In the text, we present tables 
showing only the second stage of the 2SLS 
results. The other models gave identical re- 
sults except where noted. 

SFR and Investment 
in Asset Accumulation 

If the yield impacts from SFR investments 
are to lead to sustainable increases in liveli- 
hoods, then one would expect to observe 

some degree of asset accumulation. The 
qualitative research found that this was 
indeed occurring for some households, but 
whether these are outlying cases (e.g., only 
those who have applied SFR on large areas) 
or whether this may describe the average 
household needs to be examined. In this sec- 
tion, we first describe the major asset hold- 
ings of households in the study and then 
analyze the effect of the SFR investments 
on changes in asset holdings. 

Table 7.2 shows some descriptive data 
on assets for households in the pilot villages 
and those outside. The included assets here 
are those deemed to be mobile and salable 
— all types of livestock but only some forms 
of physical capital. 25 For instance, farm 
implements were valued, but not wells. It 
can be readily observed that asset values are 
higher for households outside the pilot vil- 
lages. Part of this undoubtedly reflects the 
actual situation because farm sizes in most of 
the non-pilot villages are larger than those 
in the pilot area and livestock densities per 
household are correspondingly higher. It 
should also be noted, however, that the 
survey instruments were slightly different 
in that the pilot village survey was made to 
conform to an earlier (2000) survey. It can 
also be noticed that the magnitude of 
changes between the baseline and follow-up 
years is much larger for the non-pilot vil- 
lages. Recall that the time between the base 
period measure and current measure is 
longer in the non-pilot data set, and because 
the general economy in Kenya has been 
poor throughout the study period, this may 
have led to steeper losses of assets. 

Looking at the actual values, livestock 
comprises about 70-80 percent of the value 
of all liquid assets. The mean total wealth 
held by households was $408 in the current 

24 We did attempt to include the base period welfare indicator as a regressor, but because of our inability to over- 
come possible measurement error and regression to the mean, these results from these models cannot be properly 

25 It is too difficult to value items like houses and wells that are not often marketed and whose quality is difficult 
to assess. 


Table 7.2 Description of household liquid assets in pilot and 
non-pilot villages 


Pilot villages 

Non-pilot villages 

Livestock wealth ($) 

Baseline mean 



Baseline median 



Current mean 



Current median 



Percentage with negative change 



Total liquid wealth ($) 

Baseline mean 



Baseline median 



Current mean 



Current median 



Percentage with negative change 



Notes: Baseline date is 2000 for the pilot villages and 1997 for the non-pilot villages and 
the current date is 2002 for the pilot villages and 200 1 for the non-pilot villages. 
Values are in constant 2002 dollars. 

year in the non-pilot villages and $236 in the 
pilot villages, whereas that of livestock was 
$302 and $178, respectively. A large number 
of households suffered through disinvest- 
ment in both livestock assets and total assets 
over the period. This is remarkably consis- 
tent in both sites, with percentages ranging 
tightly between 47 percent and 54 percent. 
Notice that the absolute and relative differ- 
ences in magnitudes between means and 
medians (for both asset measures and both 
sites) are much larger in the baseline year 
than in the current year. This indicates that in 
general, households with higher initial wealth 
fared poorly compared to the less wealthy. 

Because livestock is vitally important 
in the calculation of assets, we explore this 
trend in more detail. The inverse relation- 
ship between initial livestock wealth and the 
change over time is, in fact, supported and 
elaborated by cross tabulations and Mests. 
The poor are in many cases able to increase 
their holdings of poultry, but not other ani- 
mals. But the main reason for the decreasing 
inequality is that some of the more wealthy 

households have seen their livestock hold- 
ings collapse. One of the reasons for the col- 
lapse is disease, and 50 percent of farmers 
in the non-pilot villages whose poultry 
holdings changed over time claim that dis- 
ease was the key factor. Another important 
factor is the forced selling for obligations. 
Funerals are very common because of AIDS 
and other diseases, and the slaughter of goats 
and cows is still followed by custom. With 
such high poverty rates, the wealthy are 
being increasingly called on to provide ani- 
mals for these occasions. 26 Thus, between 
26 and 33 percent of changes in goat or cat- 
tle holdings are attributed to forced sales/ 
slaughters. Given these sobering trends, the 
importance of protecting or building assets 
from SFR investments or any other inter- 
vention cannot be overstated. 

We now turn to the econometric analysis 
undertaken to test whether the agroforestry- 
based SFR technologies have had an impact 
on asset portfolios. Livestock and total 
wealth followed similar patterns because it 
was mainly livestock wealth that changed 

6 It is not known whether livestock owners were compensated for providing their animals. 


Table 7.3 Econometric results from second-stage regression of agroforestry on 
changes in assets in pilot villages (// = 97) 


least squares 


Coefficient estimate 

Significance level 

Predicted area under fallow 



Predicted area under biomass transfer 



Luo ethnic group 



Female-headed household 



Household head obtained primary education 



Household head obtained secondary education 



Household head age 



Household size 



Farm size 






R 2 


Probability of F 


p-values: "significant at 5 percent level or less; *significant at between 5 and 10 percent levels. 

over time, so we examine only one depen- 
dent variable, the change in total asset wealth. 
Table 7.3 shows the results from a sec- 
ond-stage regression of agroforestry adop- 
tion on liquid asset change. Neither of the 
agroforestry variables is significant. In fact, 
the only significant variable is farm size; it is 
found that asset holding positions changed in 
more positive directions where farm sizes 
were smaller. This suggests that non-land 
assets are not highly correlated with land 
assets, a reflection of market imperfections 
in land relative to other assets. The general 
lack of significance among other variables 
indicates the existence of complex relation- 
ships that are not easily captured by more 
structural household variables. 

SFR and Expenditures 

In this section, we examine the effect of the 
use of the SFR technologies on household 
expenditures. We begin with a brief descrip- 
tion of expenditures and then follow up 
with econometric analysis. The econometric 

analysis uses the same approach as done in 
the analysis of assets and thus we do not re- 
peat the exposition of the methodological 
approach. Expenditures were collected for 
the pilot village subsample of 103 house- 
holds both in 1999-2000 and in 2002. The 
April 2000 survey matches exactly the time 
period of the 2002 resurvey and thus we 
report on and examine only the expendi- 
tures reported at these two visits. Expen- 
ditures were collected on all types of budget 
items, including clothing, utensils, other 
household goods, fuelwood/energy, trans- 
portation, medicines and medical treatment, 
and school fees. Expenditure information 
was estimated for the preceding three-month 
period to match the baseline methodology 
that was planned as a quarterly instrument. 
In the baseline, we attempted to obtain in- 
formation about expenditures on food items 
as well. In the follow-up, this information 
was inadvertently left off and when discov- 
ered, it was then decided to omit the data for 
several reasons. 27 

27 First, the three-month recall caused households to struggle with many items in terms of amounts consumed, the 
proportion of consumption that was purchased, and the amount spent. Thus, the reliability of such data is low. 
Second, interpretation of such expenditures is complicated because food expenditure is not clearly linked to wel- 
fare, since on-farm production may substitute for it. 


Table 7.4 Total non-food expenditures, per capita non-food expenditures, and changes 
during the three-month-long rainy season in 2000 and 2002 (in U.S. dollars) 




Percent negative 

Total non-food expenditures in 2000 
Total non-food expenditures in 2002 
Per capita non-food expenditures in 2000 
Per capita non-food expenditures in 2002 
Change in total non-food expenditures 
Change in per capita non-food expenditures 














In this section, we therefore focus on the 
non-food expenditures (food consumption 
analysis follows in the next section). We 
analyzed changes in non-food expenditures 
per household and also per capita. For the 
latter we divided by the number of house- 
hold members rather than converting all 
members to consumer equivalents, as con- 
sumption coefficients are not necessarily 
constant across type of expense. For both 
the descriptive and econometric analysis, 
we examine both total and per capita expen- 
ditures. As before, two different methods of 
econometric analysis were used (pure 2SLS 
and a tobit/linear combination) but only the 
results from the 2SLS are reported, since 
the results are not qualitatively different be- 
tween the two. Table 7.4 describes non-food 
expenditures, non-food per capita expen- 
ditures, and changes in these variables be- 
tween 2000 and 2002 for 103 households in 
the pilot villages. Mean non-food expendi- 
tures in 2000 were $97, while the median 
was $60, indicating that there are relatively 
wealthy households bringing up the mean. 
The mean level of non-food expenditures 
rose slightly to $104 over the period, and 
the median behaved similarly over time. Per 
capita non-food expenditures, on the other 
hand, were flat over time, with a mean and 
median of $16 and $10, respectively. Taken 
together, the two variables indicate weak 
improvement or stagnation, which compares 
favorably with all other welfare indicators 
examined in this chapter (note that if infla- 
tion were considered, trends in real expendi- 
tures would also be negative). The final col- 

umn of the table, however, shows that there 
is also a large number of households (44^18 
percent) experiencing a setback in welfare 
as measured by non-food expenditures. 

Given the wide variation in the distribu- 
tion of non-food expenditures and per capita 
non-food expenditures, these welfare indi- 
cators lend themselves rather well to the test- 
ing of the effect of SFR technologies. Table 
7.5 presents the second-stage results from a 
2SLS procedure. The two agroforestry vari- 
ables have the opposite sign. The coefficient 
estimates for the fallow variables are nega- 
tive and significant at around the 5 percent 
level (although in the tobit model, the sig- 
nificance level is somewhat reduced after 
bootstrapping). Why this is the case is not 
apparent, because many households with 
improved fallows report real yield gains. 
Users of fallows may be driven by a few 
other alternatives and thus the use of fallow 
may be capturing effects of omitted variables 
on changes in other livelihood strategies. 
Yet, why it occurs for expenditures and not 
for any other indicator is not clear (although 
the 2SLS tests suggest that possible biases 
might arise in these expenditure models). 
None of the other included variables in 
Table 7.4 were significant. 

SFR and Food Consumption 

Food consumption and nutritional measures 
were based upon 24-hour recall surveys of 
households. During the baseline, households 
were visited on three consecutive days, but it 
was found that reducing interview schedules 


Table 7.5 Econometric results from second-stage regressions of agroforestry on 
changes in non-food expenditures and per capita non-food expenditures in pilot villages 
(/?= 102) 


Changes in nonfood Changes in nonfood 

expenditures per household expenditures per capita 

Predicted improved fallow area 

Predicted number of seasons with biomass transfer 

Luo ethnic group 

Female-headed household 

Household head obtained primary education 

Household head obtained secondary education 

Household head age 

Household size 

Farm size 


R 2 

Probability of F 













































Notes: p-values in parentheses; 

*significant at 5 percent level or less; *significant at between 5 and 10 percent 

to two consecutive days led to nearly the 
same information. So in the 2002 survey, 
households were visited only twice. 28 Enu- 
merators recorded all food consumed at each 
meal, including units, weights, and whether 
the food was consumed raw or cooked. 
They also recorded the number, gender, 
and age of all people who shared the meal. 
Recordings of consumption were not made 
at an individual level, although there is little 
doubt that consumption will differ across 
individual members. 29 Instead, household- 
level indicators of intake and nutrition were 

calculated based on age requirements of 
consuming members. Nutritional indicators 
were taken from Food and Agriculture 
Organization of the United Nations (FAO) 
and U.S. Department of Agriculture (USDA) 
sources, depending on which was able to 
more accurately reflect the specific type of 
food consumed (e.g., cooked kales). Total 
amounts of intake and nutrients were re- 
corded and an index of sufficiency was cal- 
culated for each day. Then the scores from 
the two days were averaged. The index 
assumes a value of 100 when household 

28 This was also desirable because we wished to retain two of the same enumerators who conducted the baseline 
and heavy rains during the period meant that the two were very stretched to complete the surveys in a reasonably 
short span of time. 

29 This was not done because it was agreed by all members of the team that such information could not be col- 
lected with any degree of accuracy. 


Table 7.6 Percentage of daily requirements of nutritional measures at the household 
level prior to a long-rain harvest 

Mean percent 

Median percent 

Percent households less than 




100 percent sufficiency 

Energy intake in 2000 




Energy intake in 2002 




Carbohydrates in 2000 




Carbohydrates in 2002 




Protein in 2000 




Protein in 2002 




Iron in 2000 




Iron in 2002 




Folic acid in 2000 




Folic acid in 2002 




Niacin in 2000 




Niacin in 2002 




Riboflavin in 2000 




Riboflavin in 2002 




intake or nutrition levels equaled the daily 
requirement as expressed by FAO/USDA. 

Table 7.6 shows the level of various 
consumption and nutrition characteristics in 
2000 and 2002. The variables are presented 
in the form of sufficiency of meeting the re- 
quirements according to the consumers for 
each meal. A value of 100 means that re- 
quirements have been met at the house- 
hold level. The table shows that the average 
household in the sample scores well in terms 
of energy, carbohydrates, iron, riboflavin, 
and niacin in both years. An analysis of 
baseline data revealed that maize accounts 
for 75 percent of total energy. The data for 
2002 show some diminished sufficiency 
in folic acid and low levels of protein suf- 
ficiency. These averages disguise an often 
large number of households with scores 
of below 100. For instance, 42 percent of 
households had less than 100 percent suffi- 
ciency in 2002 in terms of energy intake, 
53 percent for folic acid, and 73 percent for 

protein. The bottom line is that there is con- 
siderable variation in sufficiency levels for 
many nutritional characteristics. It is also 
interesting to note that there is a general de- 
cline in nutritional status over the two-year 
period — in fact, none of the variables ex- 
hibits improvement over time. 

Econometric analyses were made to es- 
timate the relationship between the use of 
agroforestry SFR and several of the nutri- 
tional indicators. We focused on those that 
exhibited significant change over time and 
are deemed to be vital to human development 
in the literature: energy, protein, and iron. 

As can be seen in Table 7.7, neither of the 
agroforestry adoption variables was found 
to be significantly related to changes in food 
intake and nutritional status. In fact, the 
only significant variable in each regression 
was gender of the household head, where 
female heads are associated with positive 
change (or less negative change) in each of 
the three indicators. 30 Note that decreases 

30 The only different result from the tobit-OLS 2SLS procedure was that Luo ethnicity was associated with poorer 
expenditure change as opposed to Luhya ethnicity. For all indicators except assets, the sign on the Luo variable 
is negative. The samples are from the same general area so that market and agroclimatic differences are minimal. 
The Luo community is known to have been affected more by HIV/AIDS than many other ethnic groups, but 
whether this also relates to the Luos of the Siaya highlands is not known. 


Table 7.7 Econometric results from second-stage regression of agroforestry use on 
nutritional measurements (// = 102) 





Predicted improved fallow area 







Predicted number of seasons with biomass transfer 







Luo ethnic group 







Female-headed household 







Household head obtained primary education 







Household head obtained secondary education 







Household head age 







Household size 







Farm size 














R 2 




Probability of F 




Notes: p-values in parentheses; 

*significant at 5 percent level or less; *significant at between 5 and 10 percent 

in sufficiency levels do not mean that the 
household is now facing nutritional insecu- 
rity (see Table 7.6). Therefore, the dynamics 
of food intake and nutritional status are 
very complex processes. They are not eas- 
ily pinned down to initial characteristics of 
households. They are likely to be related 
to a myriad of decisions and livelihood 
changes that take place during the period. 

SFR and the Beneficiaries 

Generally, adoption is pervaded with on- 
going social processes and the success of 
SFR technologies is then dependent on the 
entire social framework within which it takes 
place. As such, who gets to benefit and why 
can be understood only within the context in 
which these technologies are disseminated 
and implemented. Some of the key points of 
differentiation therefore include people's 
resource base, access to markets and public 

services, the type of livelihood strategies that 
they choose to pursue, the nature of vulner- 
abilities facing them, the likelihood that 
these risks can be easily mitigated, and the 
gender power relations governing their so- 
cial system. 

The four case studies narrated in Chap- 
ter 5 and this one suggest that social net- 
works are extremely crucial to one's ability 
to derive benefits from SFR technologies. 
For instance, three of these cases only got 
to know about SFR from friends and 
neighbors who were already enjoying the 
benefits. Besides being able to transfer the 
knowledge and skill required, such associa- 
tion was testimony to the potential benefits 
and a driving force behind the decision to 
take up SFR technologies. This therefore 
means that in cases where people may be 
sidelined as reported by Maria, there is still 
the possibility of benefiting through a third 


In all the cases cited, access to the three 
main factors of production, namely land, 
labor, and capital, comes through as an im- 
portant precondition to drawing benefits 
from SFR technologies. Hence, although 
Maria is relatively young and energetic, the 
amount of land at her disposal does not allow 
her to practice all the technologies avail- 
able. This is also the case for Gilbert, who 
is even double-burdened because of his 
advanced age. In addition, both of these 
farmers indicate that because of family labor 
shortages and inability to hire any, they are 
unable to carry out some of the recom- 
mended activities. In such cases, biomass 
transfer becomes one of the least attractive 
of the SFR technologies. On the other hand, 
fallow crops require that some land be set 
aside and no matter the benefits accruing, 
this is a great sacrifice for most farmers. 

Similarly, households that have diversi- 
fied their sources of income cope better with 
some of the demands of implementing SFR 
technologies. For instance, Sufu is able to 
select and even experiment with very spe- 
cific crops and technologies because he has 
a relatively better capital base (see Chapter 
5) as compared to Gilbert who, in spite of 
wanting to use farmyard manure, cannot do 
so because he has only one animal. 

The general indication therefore is that 
not everybody who has been reached in 
terms of disseminating the SFR technolo- 
gies has adopted or benefited. Some of the 
people left out are women who have found 
it difficult to participate because they lack 
sufficient land or they cannot attend the 
demonstrations/field days owing to restric- 
tions from their spouses or their heavy do- 
mestic workloads. Nevertheless, there are 
examples of households in which husbands 
and wives cooperate in the conduct of their 
farm activities with great success. In other 
words, although there is evidence to show 
that adoption and continued implementation 
of the various SFR technologies may have 
enhanced gender disparities, it is also the 
case that these technologies have managed 
to improve availability of cash incomes. 

However, as would be expected, the direc- 
tion that is taken when it comes to sharing 
these resources depends on what else is 
going on in each particular household. 

Improved fallows and biomass transfer, 
used by many poor households (see Chap- 
ter 6), are found to have perceptible effects 
on crop yields and production levels, but no 
significant impact on household expendi- 
tures, consumption, or asset building. This 
is not to say that the technologies have had 
no impact whatsoever — the qualitative and 
quantitative data indicate that a few house- 
holds have indeed benefited greatly from 
the technologies and have increased their 
welfare. However, this is confined to rela- 
tively few households. For the average 
household, the quantitative analysis finds 
that current use patterns are too modest to 
lead to appreciable impacts at the house- 
hold level. 

The amount of land or number of house- 
hold members is not an important determi- 
nant of changes to welfare indicators during 
the time period studied (an exception is with 
respect to assets in the non-pilot villages). 
The reasons for this are potentially numer- 
ous, with both agricultural and non-agricul- 
tural explanations. What is important to 
emphasize from the lack of relationship is 
that households with many different resource 
portfolios are equally likely to be improving 
or worsening in terms of a host of welfare 
indicators. It is not the case that certain 
levels of owned land or household labor are 
requirements for improved livelihood out- 
comes. Rather, households have apparently 
been able to cope when they are lacking in 
these resources and equally others have been 
unable to take advantage of relatively abun- 
dant resources. Along these lines, it should 
also be noted that gender of household head 
was often not linked to changes in welfare 
indicators, although nutritional status was 
found to be higher in female-headed house- 
holds. The results also have implications 
for poverty dynamics. While some house- 
holds are mired in low-level productivity- 
investment cycles, we found considerable 


evidence that the many households can move 
across poverty/welfare categories. Simple 
analyses of welfare change found that 
better-off households in the baseline pe- 
riod incurred significant losses in assets, 
and sometimes in expenditures or consump- 
tion. Combining this evidence with the his- 
torical case studies leads to the conclusion 
that the majority of households are vulner- 
able to poverty and targeting to a narrow 
population may be unnecessary. 

So what is driving the changes in wel- 
fare indicators? As a step toward resolving 
these complexities, we tested for associations 
between different types of livelihoods and 
welfare indicators. Participation in casual or 
salaried labor was not related to changes in 
any of the indicators, nor was the presence 
of remittances. The only relationship found 
was that households who were engaged in 
business had more positive changes in non- 
food expenditures and non-food per capita 
expenditures. Thus, like agroforestry inter- 
ventions, none of these non-farming liveli- 
hood strategies seems to contribute signifi- 
cantly to improved livelihood outcomes by 
themselves. The real factors behind changes 
in livelihood outcomes are multifaceted and 
varied across different contexts. Certainly, 
this suggests that finding causal relationships 
between contemporaneous endogenous pro- 
cesses will be almost impossible to achieve, 
given the reality of a large number of chang- 
ing factors beyond the control, measure- 
ment, and analytical tools of researchers. 

SFR and the Poor 

The SFR technologies aim at improving soil 
fertility with the ultimate goal of making 
farm practices more sustainable and profit- 
able. However, whether these technologies 
have been successful in targeting the poor 
is dependent on who is classified as poor in 
the context of the study area vis-a-vis who 
is actually capable of implementing these 
practices and in a profitable way. 

Generally, both biomass transfer and cul- 
tivation of fallow crops are best among 

smallholder farmers, most of whom engage 
in subsistence production and could there- 
fore be classified as poor. But this particu- 
lar category of rural dwellers is subject to 
various vulnerabilities, many of which are 
beyond their control. Even if they do have 
a surplus for sale, they face several odds, 
including lack of markets and market in- 
formation, poor and noncompetitive prices, 
and a relatively small amount of negotiating 
power, both at the economic and political 

Furthermore, the possibility that the 
SFR technologies being promoted will suc- 
ceed in involving the most destitute house- 
holds becomes remote with a reduction in 
the farmers' resource base through subdivi- 
sion of land, through HIV/ AIDS and other 
illnesses, and through increasing demands 
on scarce capital. Therefore, whereas the 
physical location of the project could be 
appropriate, the requirements of the tech- 
nologies may not always accommodate 
farmers who are absolutely poor. Moreover, 
even in situations where the poor manage to 
meet the basic requirements, this is neither 
sustainable nor adequate to make the ex- 
pected difference. 

Generally, most households interviewed 
pursued more than one livelihood strategy, 
mostly in mitigation against several vulner- 
abilities. In some instances, these strategies 
are employed simultaneously, while for some 
people, they are taken up sequentially. No 
matter the approach, all households inter- 
viewed were found to be pursuing strategies 
other than SFR technologies with the aim of 
enhancing their incomes. Some of these 
strategies are complementary to agriculture, 
while others run parallel or even work in 

All farmers were found to grow more 
than one crop so as to meet their diverse 
needs and as a way of spreading risk. In- 
deed, in spite of perceived dangers associ- 
ated with inorganic fertilizers and coupled 
with the fact that most of these fertilizers 
are unaffordable, a few farmers still went 
in for these inputs. Hence, whenever they 


received assistance such as remittances, 
some of the farmers were found to purchase 
inorganic fertilizers such as DAP with the 
aim of boasting their production, especially 
in cases where they are interested in im- 
mediate results. Off-farm employment and 
trading were found to be the most common 
livelihood strategies after farming. Some 
farmers, such as Sufu, opted out of farming 
to take up off-farm employment, while others 
combined these two activities simultane- 
ously. In either case the aim was to enhance 
household incomes and hopefully stave off 
poverty. And in the case of Hellena, she 
opted to sell thowi after realizing that her 
injuries could not allow her to engage in 

Social networks and enlisting the sup- 
port of others are also other strategies that 
households employ from time to time. In 
the case of Hellena, her eldest son's children 
are dependent on her for support and it is 
apparent that in the absence of the kind of 
provisions that she makes, they would be- 
come destitute. But, Hellena, too, is depen- 
dent on her daughter-in-law for labor and a 
younger son for cash income. In other words, 
therefore, when resources are scarce or 
shrouded in uncertainty, one can still keep 
out of poverty, at least in the short run, by 
depending on the good will of others. 


Several conclusions can be emphasized 
from the analysis of this chapter: 

• There is stagnation or decline in wel- 
fare indicators. 

• There are many adverse shocks that 
affect strategies of poverty alleviation. 

• The social/economic environment is 
unfavorable for rural development. 

• SFR is reaching the poor and they 
receive some benefit. 

• The benefit is not large enough to be 
visible for the majority of practicing 

• There does not appear to be any single 
rural-based livelihood strategy that 

effectively reduces poverty for many 

We found that few households exhibited 
positive changes in welfare indicators. Many 
households saw their liquid asset portfolios 
decline and their expenditure or consump- 
tion levels to stagnate or fall. It appears that 
no household is immune to welfare or liveli- 
hood losses as, in many cases, it was the 
more well-to-do households that suffered the 
greatest absolute losses. To be sure, some 
gains were made, and it should be stressed 
that almost all households are actively pur- 
suing some strategies for advancement. 

Significantly affecting the changes in 
welfare indicators are adverse shocks cou- 
pled with households' inability to cope with 
them. It was rare indeed for any of the case 
study households to be free from major 
calamity or catastrophe during the past few 
years. This includes death or severe illness 
of family members, loss of employment or 
income source, loss of livestock from dis- 
ease, unplanned contributions to funerals, 
and theft of crops. These occurrences lead 
to significant shifts in asset portfolios or 
planned investment levels. 

What these conclusions further point to 
is that the overall rural economy has been 
very weak. Investments of any kind cannot 
be effective in an environment where agri- 
culture is not rewarded (economically, cul- 
turally, socially), institutions are collapsing 
because of bad governance, and few non- 
agricultural employment opportunities are 
created for rural dwellers. Such an environ- 
ment acts to increase risks of adverse 
shocks (e.g., thefts, loss of employment) as 
well as to reduce households' ability to cope 
with shocks (e.g., risks of building livestock 

The agroforestry technologies, which 
were found to be used by a good number 
of poor households, had mixed effects on 
livelihoods and welfare. On the one hand, 
the preponderance of evidence from the sur- 
veys, the case studies, and the trials, showed 
that agroforestry did generate increased 


yields for most households. Some house- 
holds, including poor households, were able 
to benefit from the use of the technologies 
and reported building up of assets and in- 
creased food consumption. This came about 
from the intended effect of the technology 
on yields, but also from sales of tree seeds 
and to improved connections with other de- 
velopment organizations. 

However, the welfare increases were 
mainly modest, owing to the small land 
sizes on which they are applied. Moreover, 
because of acute need for cash, such in- 
creases were not usually converted into 
sustained increases in livelihood assets. The 
state of the local economy and starting asset 
bases of households play a large role in this. 
Poor asset portfolios mean that households 
cannot fully take advantage of the new tech- 
nologies in the sense of taking more risk 
with larger areas or complementing them 
with improved varieties or new crops. The 
weak economy means that there are few off- 

farm livelihood sources that could diversify 
risks and provide working capital for farm 

Given that welfare of most households 
declined, there do not appear to be any sin- 
gle promising livelihood strategies to escape 
poverty. Households with relatively large 
livestock numbers suffered larger decreases 
in welfare than others. Larger farms were 
not able to do better than smaller farms. 
Other forms of soil fertility investment (e.g., 
fertilizer) were not highly accessible to the 
poor. Moreover, none of the non-agricultural 
livelihoods were strongly linked to improved 
welfare, although in some cases, it was clear 
that non-agricultural foundations were very 
important. There is clearly no sure way to 
climb and remain out of poverty — it seems 
to require extreme flexibility and adaptation 
by households. Diversification from subsis- 
tence crop cultivation is imperative for all of 
the poor, but the direction that they take will 
not be uniform. 


Dissemination of Soil Fertility Replenishment 
Technologies: Comparing Approaches, 
Methods, and Experiences 

While much of the study focuses on impacts of the technology itself, it also looked 
at dissemination processes. A study of dissemination is important to this study be- 
cause (1) dissemination approaches used by organizations in western Kenya are 
intended not only to disseminate technology, but to strengthen human and social capital such 
that farmers and farmer groups are able to disseminate technologies to other farmers in the 
village and ultimately to other villages; (2) approaches to dissemination, methods, and expe- 
riences affect these organizations' ability to reach the poor and women — in other words, the 
process of dissemination can have as much impact on adoption as the nature of the technol- 
ogy itself. It is thus important to understand the different approaches used by different organ- 
izations in western Kenya, the experience of implementation in practice, and the effectiveness 
in achieving the objectives identified in (1) and (2) above. 

More specifically, the next two chapters address the following issues: the extent to which 
dissemination approaches have reached different groups of farmers; people's perceptions of 
the disseminating organizations and methods of teaching; the relationships between the dis- 
seminating organizations and communities; flows of information between different types of 
disseminating and recipient institutions and individuals; the effectiveness of the training; 
perceptions of local sources of information; the performance of farmers and local groups in 
carrying out dissemination; the effectiveness of the approaches in reaching the poor and 
women; impacts on human capital, social capital, and local social relationships; and sustain- 
ability issues. More detailed accounts of experiences in each of the six focus group villages 
are found in Appendix B. 

Study Design and Methods 

The methods used to design the research questions and criteria used for site selection are dis- 
cussed in Chapter 2. To reiterate briefly, here, the research took place in a selection of villages 
that overlap with villages selected for the survey and household case studies. The main meth- 
ods used for studying dissemination were focus groups and a household survey, supplemented 
by some case study material. Twenty-four focus groups were conducted across six villages 
(three Luo and three Luhya), with four groups per village representing poor and non-poor 



Table 8.1 Village selection for dissemination study 

Type of 





Focus group 


Luo villages 


West Kanyaluo 

Luhya villages 

Village approach 
TRACE approach 
Village approach 
Catchment area approach 
Sub-chief visited Maseno 
Local CBO through ICRAF 

Village approach 

Catchment area approach 


Umbrella group approach 











Local leaders 














a The main disseminating organization is in bold. The rest joined in after the approach was in place and used it to reach farmers. Note that other 
organizations were also active with projects that may have included soil fertility in some of these villages, but were not the main SFR inter- 
ventions of interest to this study. Acronyms not defined earlier: KWAP, Kenya Woodfuel Agroforestry Programme; TRACE, Training of Re- 
source Persons in Agriculture for Community Extension. 

men, and poor and non-poor women. 31 The 
survey covered 360 households across six 
villages (three Luo and three Luhya), but 
only four overlap with the qualitative study 
on dissemination. 32 Table 8.1 shows this 

Each focus group used two types of 
information collection methods. One was 
open-ended discussions, using a structured 
guide but allowing responses to take their 
own directions and participants to discuss or 
debate with each other. The second involved 
the use of several PRA methods to allow 
people to visually express their evaluations 
of dissemination organizations and pro- 

cesses and then debate the responses. These 
included the methods described below. 

Mapping of Organizations 
and Information Flows 

The purpose of the exercise was to learn 
which local and external organizations and 
individuals were operating in the village 
and involved with providing, passing on, or 
receiving information and training about 
soil fertility replenishment and agroforestry 
methods, and the means and media through 
which information was passed on. It aimed 
to understand how people accessed informa- 

3 'In the following two chapters, "poor men" refers to participants in the poor men's focus group, "non-poor 
women" refers to the focus group participants in the non-poor women's focus group, and so on. It does not, of 
course, mean that they speak for everyone inside or outside the group, although information included in these 
chapters tends to reflect a response that was prevalent within the respective focus group. Furthermore, "poor" and 
"non-poor" are relative categories: even the non-poor are often households struggling to make ends meet, and 
many might be considered poor in relation to an urban working class household, or compared to some categories 
of rural dwellers. Selections were made based on survey data described in this report, so definitions of poor and 
non-poor can be found there. 

32 The quantitative study of dissemination, adoption, and impact was administered in non-pilot villages only, so 
Sauri pilot village was not in the sampling frame. Second, Gongo was not part of the quantitative study because 
too few households were found to be practicing agroforestry, and Ugunja was selected in its place. 


Figure 8.1 Village map of institutions involved with SFR information exchange— Bukhalalire village, poor men 

Field day 



Exchange of ideas 

-► Strong both ways 
— Strong one way 
Weak both ways 
Weak one way 

tion, how information moved through the 
community, and where barriers or weak 
linkages existed that prevented access or 
made access unreliable. 

A map was drawn on a large sheet of 
paper on the ground, representing the bound- 
ary of the village. Farmers were asked about 
all providers of information on soil fertility 
management, agroforestry, and conservation 
from outside the village, and to place these 
outside the village map. These included, for 
example, ICRAF, Ministry of Agriculture 
and Rural Development (MoARD), and 
CARE. They were then asked about pro- 
viders of information within the village, and 
these were placed inside the map. Some of 
these included local women's groups, vil- 
lage committees, schools, other farmers, and 

contacts farmers. Some organizations were 
placed at the boundary. Farmers were then 
asked to list the medium of information ex- 
change between the providers of informa- 
tion and themselves. Each of these was 
drawn in below the provider that uses them. 
Some of these included barazas, field days, 
farmer exchanges, demonstrations, and tours. 
Finally, the facilitator asked farmers about 
the linkages between the organizations, 
groups, and individuals within the village, 
between external organizations, and last 
between those on the inside and the out- 
side. Participants drew these linkages, rep- 
resenting strong or weak communication, 
and whether the communication was in one 
direction or two-way. Examples of the draw- 
ings can be seen in Figures 8.1-8.3. 


Figure 8.2 Village map of institutions involved with SFR information exchange— Mutsulio village, poor women 


Visiting farmers 

Visiting farmers in 

Baraza meeting 

Visiting farmers 

Tours and 

Visiting during funerals 
Normal visits 

•^ ► Strong both ways 

•4 Strong one way 

■^ ► Weak both ways 

■4 Weak one way 

Scoring of External Information 
Providers and Methods 

Having established what information pro- 
viders operated in the village and the media 
of information, farmers were then asked 
to score the usefulness/importance of each 
provider and media. This exercise was done 
several times: for external providers, inter- 
nal providers, and methods of dissemina- 
tion. Criteria for evaluating the usefulness 
and importance of particular providers of 
information were solicited from the group. 
For each of the three categories, participants 
were given 100 grains of maize (or beans) 
and asked to give each provider a score of 
between 1 and 100, with decisions about 
relative importance represented by the num- 
ber of grains. The total number distributed 
was limited to 100, so that results could be 
expressed quantitatively in percentages. In 
all cases, people were asked to express their 

reasons behind the scoring and their answers 
were probed. 

Knowledge Acquisition: Ladders 

This exercise focused on the effectiveness 
of training and farmer self-evaluations of 
their knowledge gained on soil fertility re- 
plenishment before, during, and since the 
intervention. The facilitator started by ask- 
ing the farmers to list all the soil fertility 
management and conservation technologies 
that the main provider of information was 
disseminating in the village. In some cases, 
if the discussions were being held in some- 
one's compound, the farmers were asked to 
bring a sample of the materials that they 
were using, for example, handfuls of com- 
post and animal manure, leaves of tithonia 
shrubs, legumes fallow species, and spoon- 
fuls of DAP. Participants were asked to grade 
their knowledge at three periods in time: 


Figure 8.3 Village map of institutions involved with SFR information exchange— Mutsulio village, poor men 






Visiting farmers 

Visiting farmers 










in farms to 

Visiting homes 

Visiting farmers 






Discussions at 

Strong both ways 
Strong one way 
Weak both ways 
Weak one way 

(1) before the intervention/provider arrived; 

(2) the period they spent with the provider; 
and (3) since the provider left. The facilita- 
tor brought out two (real) ladders with 10 
rungs each and explained that each rung 
represents an increase or decrease of 10 per- 
cent, starting with percent (no knowledge 
concerning a technology) and ending with 
100 percent (total knowledge in terms of im- 
plementation and ability to pass on to others). 

For each technology, farmers indicated 
where on the ladder they were before, dur- 
ing, and since the intervention. If they had 
samples of the materials, these were placed 
on the rungs physically using tape. Discus- 
sions, arguments, and reasons for grading 
each level were captured. 

Quantitative Survey 

A sample of 360 households, 60 from each 
of six villages, was selected. Of these, ap- 
proximately half were using agroforestry- 
based SFR technologies. The survey instru- 

ment was broad, covering basic household 
characteristics, acquisition of information 
on agroforestry, use of agroforestry, and 
impacts from agroforestry. Among those 
related to acquisition of information were 
questions about the household's awareness 
of different disseminators of information 
about SFR, their direct interaction with the 
disseminators of information, and their as- 
sessment of the frequency and reliability of 
the information received. We also asked 
questions to better understand how house- 
holds were connected in the community — 
their positions of leadership and their mem- 
bership in groups. 

Approaches to 
Dissemination in 
Western Kenya 

As Table 8.1 shows, a range of different 
organizations are disseminating SFR tech- 
nology approaches used in this region, 


and the organizations use different systems 
for dissemination, which we refer to as "dis- 
semination approaches." These share cer- 
tain methods of organizing and teaching, 
but differ in a number of ways. Below are 
descriptions of the approaches used in the 
six villages in which this research took place. 
The actual experience may vary in practice, 
or people's perceptions may be different 
from what the organizations intend, as evi- 
dent in differences between the descriptions 
below and the study findings that follow. 
Nevertheless, we describe the approach as 
designed by the disseminating organization, 
and refer to the intended or actual experi- 
ence across the wider group of villages in 
which the approach was used. 

Village Approach— ICRAF 
Dissemination Approach 

ICRAF and its partners (KARI and KEFRI) 
initiated a "village approach" to dissemina- 
tion. Initial contact was made through the 
assistant chief and village headman where 
they discussed soil fertility problems and 
possible solutions. The village headman in 
turn organized a field meeting for all farm- 
ers in the village to meet with researchers. 
Everyone was exposed to the technologies 
and each farmer tested at least one strat- 
egy. With connectivity between villages, re- 
searchers decided to work with the cluster 
of villages within the sub-location, provid- 
ing the opportunity to include a wide range 
of farmers from different socioeconomic 
and cultural groups. It was necessary to 
develop a community-based extension sys- 
tem that would facilitate the dissemination 
of some of the new locally developed tech- 
nologies in a sustainable manner. The ap- 
proach involved institutionalization of ex- 
isting community social structures. Using 
participatory methods, social and clan groups 
in each village were identified and linkages 
among the groups in terms of membership 
were established. Each group and/or clan 
group selected a representative to be part of 
the village committee, responsible for pro- 
viding information to group members. After 

all the villages had formed their committees, 
each village committee (VC) elected two 
representatives to form the sub-locational 
committee (SLC). It was at the sub-location 
level that information from external agents 
could enter the community and spread down 
to each farmer through the VC and groups. 
The various committees were trained in 
technical agroforestry aspects of seed pro- 
duction, handling and storage, nursery es- 
tablishment and management, soil fertility 
replenishment strategies, group dynamics 
and team building, record keeping, leader- 
ship skills, monitoring and evaluation, and 
proposal writing. Farmers were reached 
through the following channels, field days, 
tours and exchange visits, seminars at the vil- 
lage and at research centers, chief's baraza, 
church services, funerals, and village meet- 
ings. After training of the committees on 
different issues mentioned earlier, it was the 
responsibility of committee members to train 
other farmers. The SLC organized field days 
for farmers from other villages to visit and 
learn what they are doing. Also during the 
chief's baraza, the chairperson of the SLC 
could talk to farmers about agroforestry 
technologies. The committees were also in- 
volved in sourcing for new technologies and 
bringing them to all farmers in the village. 
The chairperson of the SLC and VC in- 
formed farmers of upcoming events such as 
field days, seminars, and meetings. Occa- 
sionally, scientists participate in village and 
SLC meetings and share ideas with farmers. 
The SLC and VC have taken up new initia- 
tives that complement the soil fertility tech- 
nologies, for example, managing an input 
credit scheme, rotational dairy cows pro- 
gram, apiculture, and growing of high- value 

Catchment Area Extension 
Approach — MoARD 

The Ministry of Agriculture and Rural 
Development is mandated to provide exten- 
sion services to farmers in Kenya. There 
are trained Frontline Extension Workers 
(FEWS) and Divisional Extension Coordi- 


nators (DECs) whose responsibility is to 
implement the approach and provide exten- 
sion services. Having begun as a soil con- 
servation program, the catchment approach 
evolved (from the 1970s to the 1990s) to 
embrace all farming issues from farm pro- 
duction, soil and water conservation, and 
marketing to farm management, home eco- 
nomics, and other activities. 33 It relies on 
creating forums where farmers play the lead- 
ing role in identifying their problems, op- 
portunities, and solutions. In one year, all 
physical and human efforts of extension are 
concentrated in one site within a division. 
Such areas/sites normally are about 400 
hectares in area comprising 400-500 farm 

The first step is focal area identifica- 
tion and selection conducted by land users/ 
farmers, FEWs, and DECs, followed by 
publicity through field baraza, churches, and 
schools. PRA is carried out, an action plan is 
drawn up in a public meeting stating activi- 
ties and responsibilities of each participant, 
and a catchment area committee is formed, 
who are trained for three days. Catchment 
area scheme planning entails developing a 
focal area map and list of farmers and/or 
households, development of community- and 
farm-specific action plans by farmers and 
extension staff, and implementation of ac- 
tion plans. Demonstrations and field days 
are then conducted according to the action 
plan, farmer tours and excursions are orga- 
nized and held as necessary, and continuous 
participatory monitoring and evaluation is 
carried out. The FEWs and DECs conduct 
field visits and prepare reports. At the end of 
the year, a five-day workshop is held to 
review work and need for maintenance and 
continued implementation of specific action 
plans. Apart from using the catchment com- 
mittee to reach all the farm families within 
the catchment, the FEWs and DEC use 

field days, social groups (e.g., women, youth, 
church, and self-help groups), chief's baraza, 
funerals, radio, and village meetings. Estab- 
lishment of demonstrations on farmers' fields 
allows farmers to learn from each other and 
exchange ideas through conducting field 
days on those farms. 

Training of Resource Persons 
in Agriculture for Community 
Extension (TRACE) 

This is a participatory extension methodol- 
ogy that is used by CARE-Kenya in its 
agroforestry program. It is based on the 
hypothesis that communities are organized 
into groups and it is easier in terms of time 
and resources to work with these already es- 
tablished groups. TRACE used groups and 
schools in a simultaneous and supplemen- 
tary manner. The TRACE process is imple- 
mented through extension and training, 
institutional capacity building, and adaptive 
research. TRACE starts with site selection 
where all the stakeholders discuss and agree 
on a specific area. Later the site (referred to 
as synergy area) is characterized and leaders 
of this area are trained. Leaders from sur- 
rounding sites are also trained. During this 
training, the Locational Development Com- 
mittee (LDC) is formed. Farmers are se- 
lected from the village level to be Village 
Agricultural Promoters (VAPs). At the same 
time, the groups in the village select repre- 
sentatives who are called Group Resource 
Persons (GRPs). The VAPs and GRPs act as 
Adaptive Research Farmers (ARF) who 
conduct trials on their farms and these farms 
are used as learning grounds. Any new tech- 
nology is first tried by the two groups of 
farmers. The VAPs and GRPs sit at the 
Village Management Committee (VMC). 
Schools in the area are part of the VMC. 
If there are many, then they form a school 
committee that regularly attends meetings 

33 As of 200 1 , this is now the official extension approach in cultivated areas of Kenya, known as the National 
Agriculture and Livestock Extension Programme. 


with the VMC. Several representatives from 
each VMC join to form a Sub-location Man- 
agement Committee (SMC). 

The overall body is the Locational De- 
velopment Committee (LDC), which is used 
as entry point into the community. The LDC 
members are chiefs of locations and they 
are trained in participatory approaches. 
This serves to eliminate bias that could be 
introduced by staff and promotes sustain- 
ability, providing an exit strategy for the 
project. The CARE groups were connected 
to the agroforestry structures formed by 
ICRAF. At the SMC, some farmers were 
selected to the Locational Agroforestry 
Committees (LACs) and the chief is the pa- 
tron. The LACs serve in raising awareness 
and community mobilization, coordination/ 
facilitation of groups, schools, and ARFs. 

The LDC and LAC members are 
trained in trial management, monitoring and 
data recording, and analysis. Groups, on the 
other hand, are trained in group dynamics, 
farm layout, and record keeping. The train- 
ing is held in a GRPs farm and field days are 
held at opportune times in the season. GRPs 
were occasionally taken for exchange visits 
to other sites. Methods used to reach farm- 
ers and pupils include field days, baraza, 
group meetings, environmental clubs, spe- 
cial class lessons, churches, exchange visits, 
evaluation meetings, and parents' days. 
Farmers were involved in committee forma- 
tion and more women were involved than 
men because the groups that were interested 
were mainly women's groups. Most of the 
GRPs are women and the VAPs are men. 
The farmers were involved in managing 
experiments, record keeping, extension 
message development, and dissemination. 
Farmers also took over organizing field days 
and village meetings. 

Participatory Learning and Action 
Research (PLAR) Approach 

Participatory Learning and Action Research 
(PLAR) is a research and extension approach 
that stimulates individual and communica- 
tive learning for Integrated Soil Fertility 

Management (ISFM) and was implemented 
KIT (Royal Dutch Institute for Tropical 
Agriculture). The PLAR approach was 
aimed at helping farmers improve their soil 
fertility management through on-farm learn- 
ing, self-discovery, and experimentation. Ini- 
tially, the PLAR team contacted the assis- 
tant chief and informed him of the activities 
to be conducted in the village within his 
sub-location. The assistant chief invited all 
farmers to an introductory meeting and the 
PLAR team attends. Several analyses are 
then done: at the individual farm level an 
analysis is made of the specific circum- 
stances of each farm and farm family, their 
problems, financial sources, access to inputs, 
their hopes, and opportunities. Resource 
flow maps are used to visualize and analyze 
farmer practices, rotation schemes, input 
supply, and other soil fertility strategies. 
Planning maps are used to plan for alter- 
native practices. Through experimentation, 
individual farmers adapt technologies to 
their own situation and integrate them into 
their farming system. At the village level, 
an analysis is done on farmers' information 
and communication networks, since infor- 
mation sharing is the key to PLAR, and of 
SFR strategies prevalent in that particular 
community setting. 

Through this analysis, farmers are 
grouped into good, average, and poor soil 
fertility managers. Farmers are selected from 
these three categories during a community 
meeting, called test farmers, and form the 
interface between farmers and external 
change agents. They form the village com- 
mittee for ISFM, and are involved in devel- 
oping options for ISFM. They are encour- 
aged to exchange experiences and views 
with their peers and the PLAR team, and are 
introduced to new information. The results 
of test farmers are regularly aired in farm- 
ers' groups and community meetings, in the 
hope that they will stimulate other farmers 
to take action. Each SFM committee draws 
a village action plan containing activities in 
the field of training, experimenting, moni- 


taring, and evaluation. The SFM committee 
organizes field days and seeks information 
for other farmers. 

With this approach, each member of the 
community has access to information and 
ownership is established through training of 
the village committee members on planning, 
monitoring, and evaluation. Interactions 
between farmers and extensionists (which 
in this case are represented significantly by 
KARI researchers) are on three levels: com- 
munity meetings, group meetings with se- 
lected farmers, and household meetings with 
members of selected farms. Farmers are 
encouraged to communicate horizontally by 
exchanging information and insights. Regu- 
lar meetings are central elements in the pro- 
cess and they rely heavily on open-ended 
conversations and regular exchange of find- 
ings from the test farmers. Since they are 
selected from groups with similar SFM 
strategies, the village committee (composed 
of test farmers) is responsible for sharing in- 
formation among farmers with similar char- 
acteristics. At the same time, they seek in- 
formation from external agents. The roles of 
extension staff in PLAR are to assist farmers 
in learning and self-discovering, experimen- 
tation and in their search for new infor- 
mation, facilitate farmer-to-farmer learning, 
stimulate interaction between committees, 
assist the functioning of farmers' platforms, 
stimulate contacts of committees with other 
development organizations, and share 
knowledge with research/extension staff of 
all levels. 

Umbrella Development Group 
Approach— KWAP 

Kenya Woodfuel Agroforestry Programme 
(KWAP) was based in Busia District in 
western Kenya. After conducting problem 
identification and analysis, farmers and the 
KWAP team chose the Mirror technique 
(see later) and the school program to be 
undertaken through the Umbrella Develop- 
ment Group approach (UDG). The UDG is 
an amalgamation of several social groups in 
an area, including women, youth, churches, 

self-help and welfare groups, and develop- 
ment societies. 

The extension approach used a frame- 
work involving "A, B, and C" areas: "A" 
was a pilot area that was a catchment com- 
posed of about 400 households; "B" was an 
administrative location in which the pilot 
area falls; and "C" was an intervention agro- 
ecological zone in which a pilot area falls. 
In A areas, UDGs were formed and KWAP 
worked intensively with partners who in- 
cluded extensionists from government agen- 
cies and NGOs operating in that area. In B 
and C areas, KWAP left the work to line 
agencies that had mandated to offer exten- 
sion services and KWAP's role was to facil- 
itate these line agencies in execution of their 
duties. In A areas, segregated farmer groups 
had existed, that is, catchment committees, 
women groups, youth groups, adult educa- 
tional groups. KWAP assisted these groups 
to consolidate into site umbrella develop- 
ment groups (UDGs), in order to have bet- 
ter bargaining power in terms of resource 
acquisition. These UDGs, which are com- 
prised of 30 members, had responsibility for 
coordinating and steering the developmental 
activities of individual groups. The UDGs 
had various subcommittees, which were 
charged with different responsibilities. One 
such subcommittee was the Adaptive Re- 
search Farmers Committee (ARFC), whose 
role was to develop and test any promising 
technology on behalf of the community. All 
the UDG members had the responsibility of 
being Resource Persons (RPs) for their re- 
spective groups in the technology transfer 
process. The whole program had six catch- 
ments in total with six to eight Adaptive 
Research farmers in each catchment. All 
resource persons had three to four follower 
farmers for closer guidance in their respec- 
tive groups. 

KWAP's role was to strengthen the 
UDGs in terms of technical and managerial 
capacities through training and farmer edu- 
cational tours. KWAP also carried the cost of 
production of the teaching materials. KWAP 
organized two main methods of teaching. 


The mirror technique employs use of a 
"mirror" in front of the community so that 
it is able to detect and understand its own 
problems. This was done through local 
drama, songs, role-plays, films and poems in 
baraza and field days, and TV productions 
to communicate agroforestry messages. This 
methodology was geared at triggering dis- 
cussions, creating awareness, and changing 
the attitude of household members toward 
tree planting (particularly woodfuel, which 
is regarded as a secondary need and as 
woman's responsibility). The mirror tech- 
nique was aimed at encouraging farmer-to- 
farmer extension (since it is farmers acting 
and/or singing). The plays and songs were 
also heard on the local radio station. 

The KWAP team also used schools 
for dissemination, where committees com- 
posed of students and teachers were formed. 
The committee organized field days, sensi- 
tized and prepared the community for the 
event, and prepared their tools for passing 
information (that is, drama, songs, poems, 
and role -plays). During these events, as many 
as 1,000 people could attend. 

Outside Institutions, 
SFR Technology, and 
the Study Villages 

As indicated in Table 8.1, the six villages in 
the qualitative study received interventions 
from several disseminating organizations, 
with one organization usually primary with 
respect to dissemination of SFR technolo- 
gies. However, because in practice organi- 
zations overlapped in their coverage, local 
residents sometimes had trouble identifying 
which outside organization was most active 
in the community, and it was rare to find 
anybody who could readily identify the name 
of the dissemination approach, for example, 
"PLAR" or the "catchment" approach. 
Nonetheless, this lack of recognition does 
not stop people from developing strong, di- 
vergent opinions on organizations. These 
differences in opinion reflect people's expe- 

rience with the organizations' dissemination 

Disseminating Organizations 

According to our study design, each of the 
six villages in the qualitative studies was 
to focus primarily on the dissemination ap- 
proach and associated external institution 
that were the most prominent with regard to 
SFR. However, in all villages more than one 
institution was working with different forms 
of SFR, and some organizations had worked 
in each village in some capacity at some 
point in time. For example, the government's 
extension service (through MoARD) in the- 
ory would work in every village regardless 
of a major intervention by another institu- 
tion, and CARE might invite ICRAF into 
villages where it is working to introduce 
new information. Hence, in each village, we 
may touch on multiple interventions where 
villagers raise them, but an effort is made to 
focus on the approach and institution that 
was particularly influential. As in much of 
western Kenya, the villages were also hosts 
to a large number of other organizations ad- 
dressing different problems, from malaria 
control to credit schemes to wells. Although 
an attempt was made not to cover these 
organizations, their presence in the villages 
means that they were sometimes compared 
in the focus groups to the main organiza- 
tions of interest. 

Of the qualitative study villages, Sauri 
and Mutsulio had four organizations active 
in SFR at some point in time, Muhanda and 
Bukhalalire had three, and Mwitubi and 
Gongo had two organizations active. Focus 
groups participants did not refer to or de- 
scribe the approaches specifically, but they 
were aware of which organizations worked 
there and the methods they used. 

Assessments of External 
Disseminators in Communities 

Respondents in the focus groups have strong 
opinions about the usefulness and impor- 
tance of the different organizations that 


worked in their villages. Five organizations 
were primary disseminators in the six study 
villages: KARI, MoARD, ICRAF, CARE, 
and KWAR Of these, residents were usually 
aware of four of them: KARI, MoARD, 
ICRAF, and CARE. KWAP was relatively 
unknown because it was only active in one 
village (Bukhalalire). Where organizations 
were not present or their work was much ear- 
lier or not as central as an SFR intervention, 
we do not assess residents' views of these 
organizations. Table 8.1 indicates which or- 
ganizations were mentioned as being familiar 
in each village, with the main organizations 
of interest to the study in bold. Figure 8.1 
presents an example (in the eyes of the 
poor men's focus group in Bukhalalire) of 
the number of organizations involved in one 
village, both external and internal to that 
community, and in what ways and how well 
information flows between them. 

Comparing the four organizations most 
active in the communities (KARI, MoARD, 
ICRAF, and CARE) seems to reveal unifor- 
mity of opinion. In fact, all organizations 
score approximately equal on the PRA exer- 
cises among those groups that rank them 
according to their usefulness and importance. 
Where identified, residents score MoARD 
and KARI at approximately 25 percent. 
Similarly, residents rank CARE at 21 percent 
for usefulness and importance, although it 
was not ranked in three communities. 
Among the four most common SFR distrib- 
utors, ICRAF scores highest at 32 percent 
(although it was not ranked in Congo and 
scored low in Mutsulio, where it was barely 
active). Comparing these numbers across 
communities is an imperfect measure be- 
cause not all organizations were operating or 
equally active in each village. However, the 
rough uniformity and the fact that the most 
active organizations usually score higher 
suggest a basic degree of satisfaction (e.g., 
Sauri, Arude, and Mutsulio scored ICRAF, 
CARE, and KARI, respectively, at more 
than 60 percent). Where it operates, ICRAF 
seems to rank consistently highest in use- 

fulness and importance compared to other 
organizations. In fact, considering its low 
score in Mutsulio, ICRAF's scores in com- 
munities where it was most active (average 
52 percent) are especially high. 

Although they were only the primary 
SFR disseminators in two communities, 
ICRAF is unusual in that it was active in 
all study communities. But MoARD was 
also active in every community and scores 
lower. Part of the reason for this relatively 
high assessment of ICRAF may lie in the 
way in which it works with groups. From 
the focus groups discussions, it appears that 
where ICRAF is active, it tended to use ex- 
isting social structures, such as church and 
women's groups and funeral societies as 
forums to disseminate their technologies. At 
least in the study villages, MoARD seemed 
to have formed new local groups to dissem- 
inate SFR technology, and these are less 
favored in these communities. In the two 
communities where MoARD was most ac- 
tive, Congo and Mwitubi, residents saw 
these groups as leading to more stratifica- 
tion, and disenfranchisement. 

But the most compelling reason for this 
ranking emerges from respondents' assess- 
ments of the outside disseminators' links 
with farmers. In the PRA exercise mapping 
lines of communication, 1 1 of the 24 focus 
groups showed ICRAF as actively involved 
with individual farmers. No other organiza- 
tion performed this well in this exercise. 
One complaint from participants was that 
external organizations did not spend enough 
time with farmers, so these lines of commu- 
nication may help to explain ICRAF's high 
scores. Also, the case studies reveal that 
people like ICRAF because of the high 
profile visitors and access to other organiza- 
tions that ICRAF provides. An additional 
trend that is to be expected is that people 
rank organizations that are most active in 
their community as more useful and impor- 
tant. Residents claimed that the prime dis- 
seminator is more useful in five of the six 
communities. In three of these cases, the 


most active organization dwarfs all others. 
Residents in Sauri, Arude, and Mutsulio 
scored ICRAF, CARE, and KARI, respec- 
tively, at higher than 60 percent. No other 
organization comes within more than 19 
points of them. In two other cases, Gongo 
and Bukhalalire, residents rank MoARD 
and KWAP, respectively, about 10 percent 
higher than other organizations. In the final 
case, however, residents rank the group most 
active there 30 percent lower than another. 
Residents of Mwitubi rank MoARD at 33 
percent, but ICRAF at 67 percent. 

It is also worth noting that one organiza- 
tion received only praise from participants. 
The Centers for Disease Control of the 
United States ran a mosquito eradication 
program in Gongo to limit malaria. While 
eliminating a terrible, endemic, and infec- 
tious disease would make any organization 
popular, and was outside the scope of other 
organizations, their popularity also stems 
from another reason more relevant to this 
study. They spent their time extensively 
training local collaborators and expressly 
showing the good that could come from 
adopting their technologies. This hands-on 
approach endeared them to the community, 
and made them popular in all groups. 

Evidence on Effectiveness of 
Dissemination Sources from the 
Quantitative Survey 

Table 8.2 displays a variety of information 
regarding recognition and contact with dif- 
ferent sources of information about SFR. 
Across the columns are the main sources of 
information on SFR (a few less important 
sources, such as the media, are omitted). The 
variables along the rows are then contrasted 
for each source. The first row simply states 
the number of villages where the particular 
source is active, because not all villages 
would be expected to be reached by each 
source. We expect that all villages would 
have known about ICRAF (meaning ICRAF 
and/or its research partners, KARI and 
KEFRI) as well as potentially receiving in- 
formation about SFR from extension and 

other farmers. In terms of NGOs or CBOs, 
while all villages may be in contact with 
at least one from each group, the number 
known to be providing some information on 
SFR is less. Thus, NGOs and CBOs actively 
provide information on SFR in only four 
and three villages, respectively. 

The second row indicates the percentage 
of households (using the total number of 
households in the villages relevant to the 
particular source) that are aware that infor- 
mation about SFR is provided by the dif- 
ferent sources of information. ICRAF ranks 
highest for awareness, as 44 percent of 
farmers were able to name ICRAF. The next 
most recognized source was CBOs at 32 
percent. NGOs and extension were nearly 
identical at 22 percent and 21 percent, re- 
spectively. Finally, other farmers were 
mentioned as sources for SFR by 18 percent 
of the households. NGOs may have received 
relatively low recognition because some 
of them maintain strong linkages with only 
a few farmers in a village. In other cases, 
households may be aware of the NGO, but 
perhaps for non-SFR-related activities. 

The third row shows the percentage of 
households that claim to have had direct 
contact with the source. By definition, this 
figure will be less than that of the previous 
variable. By dividing the direct contact vari- 
able by the awareness variable (that is, di- 
viding row 3 by row 2), one can generate a 
measure of direct support provided by each 
source (in parentheses in row 3). In these 
terms, CBOs score highly in that they have 
direct contact with over 90 percent of those 
households who are aware that they provide 
information on SFR. Farmer-to-farmer meth- 
ods of dissemination also tend to involve 
personal contact in most cases. Extension 
agents also seem to be effective in reaching 
clients, at least those who are aware of ex- 
tension as a source for SFR information. 
NGOs and ICRAF, the two external organi- 
zations, are found to have the least perva- 
sive direct support to households. This is 
not surprising, given that they work exten- 
sively in many different villages. 


Table 8.2 Summary of reach and effectiveness of different sources of information on 
SFR (as percentage of all households located in relevant villages) 

Indicator CBOs NGOs Extension Farmers research 

Number of relevant villages 

Percentage of farmers in these villages who mention 31.7 21.8 

them as source of information on SFR 

Percentage of farmers in these villages with direct 
contact on SFR 

Percentage of farmers in these villages satisfied 
with the information on SFR 































Notes: In row 3, italicized numbers represent percentage of households having direct contact among those aware 
that the source provides information on SFR (row 3 divided by row 2). In row 4, italicized numbers rep- 
resent percentage of households satisfied with information among those with direct contact (row 4 divided 
by row 3). 

The final row shows the percentage of 
households reporting to be satisfied with the 
SFR information received from the source. 
Dividing this percentage by the percentage 
of households with direct contact (in the 
parentheses in row 4) provides a measure 
of the effectiveness of the information pro- 
vided by each source. 34 On this account, 
CBOs and NGOs rate highly, with approxi- 
mately two thirds of contacted households 
being satisfied with the information pro- 
vided. Other farmers provide satisfactory 
information in about half the cases and 
ICRAF in just about 40 percent. The lowest 
rated source of information is extension, 
which does not provide adequate informa- 
tion in about 75 percent of cases. It is pos- 
sible that poor extension services in other 
aspects of farm production may have led to 
poor ratings for SFR. 

It is not possible to directly infer from 
this information the effectiveness of spe- 
cific approaches. For example, the fact that 
a source of information scores relatively 
low on direct contact may simply result 
from the fact that they intended to have a 

low direct contact and planned that contact 
farmers would in turn pass on the informa- 
tion. Eventually, one would wish to know 
how many households in total have heard 
about SFR. One vehicle that is encouraged 
by all approaches is farmer-to-farmer con- 
tact. Table 8.3 shows the percentage of 
households who received information about 
SFR from fellow farmers in the different 
sites. As can be seen, farmer-to-farmer ex- 
change is highest in Mwitubi (26 percent 
of households where MoARD and ICRAF 
were quite active) and then in Muhande 
(39 percent of households where CARE 
was active for many years). Buhkalarire was 
lowest with only 5 percent, having received 
information from other farmers, but as 
noted earlier, this may be partly the result 
of very active dissemination by local 
groups. In terms of access to SFR from any 
source, 73 percent of households across the 
six sites replied that they did receive some 
information. This implies that information 
dissemination is very good overall in these 
early dissemination sites. The two sites 
that are relatively low are Buhkalarire (63 

34 One could also divide by the awareness percentage, but it is not clear that households that are aware of a source 
of information actually received information with which to make an assessment of quality. 


Table 8.3 Percentage of households receiving information on agroforestry from other 
farmers or any source 


Percentage of households receiving 
information from other farmers 

Percentage of households receiving 
information from any source 






West Kanyaluo 



percent) and West Kanyaluo (57 percent). 
These happened to be the furthest sites 
from ICRAF's base in western Kenya and 
in the case of West Kanyaluo, recall that 
these farmers visited Maseno for infor- 
mation and no organization was present 
in the village to disseminate information 
on SFR. 

Differentiation within 
Communities in the 
Focus Group Results 

Returning to the focus group results, some 
differences were found between men and 
women, and between non-poor and poor 
focus group participants, in their attitudes 
toward disseminating organizations. In the 
aggregate, the average scores for men and 
women were roughly similar. Three of the 
four outside groups get almost exactly equal 
scores from both men and women at 30 
percent. The only significant difference is 
for ICRAF. Here, men rank ICRAF at 50 
percent, while women rank it at slightly less 
than 40 percent. This is a small difference 
and difficult to explain. ICRAF's use of ex- 
tension approaches that visit farmers in the 
field could have gained them popularity with 
men who have greater visibility as head of 
household and greater access to land. How- 
ever, ICRAF does attempt to reach women 
farmers, and is successful as evidenced by 
the high score they do receive from women. 
In fact, direct visits with farmers in their 
fields can also favor women who might be 
less inclined to attend public dissemination 

Viewing results by wealth, there are 
some significant differences. Three of the 
five organizations (CARE, KWAP, and 
KARI) are more popular among non-poor 
residents. For example, non-poor groups 
rank CARE as about 30 percent more im- 
portant than do the poor groups. In con- 
trast, MoARD and ICRAF are more popu- 
lar among poor residents. These distinctions 
suggest that these two organizations are 
reaching poor farmers. Since focus group 
discussions indicate that, generally, poor 
farmers have less positive experiences in 
groups, they might appreciate ICRAF and 
MoARD's practice of visits to individual 
farmers. Although this was popular every- 
where, poor groups, and especially poor 
men, unanimously liked this method. 

But other patterns emerge in combining 
gender and wealth. For example, KARI and 
KWAP are both more popular among poor 
women than non-poor women, but more 
popular among non-poor than poor men. 
One possible explanation for this pattern 
could lie in the heavy use of local groups 
in their approaches. In both communities 
where these groups were most active, all 
men, but especially poor men, noted that the 
groups founded by disseminators created 
stratification, since they relied on local elites 
as their contacts. In contrast, both groups of 
women, but especially poor women, said 
that these dissemination groups "eliminated 
social differences." The elite men in these 
groups may have dominated less well-off 
men, while not affecting women's groups, 
where they were not present. 


In contrast, poor men and non-poor 
women assess MoARD most positively. 
This is of particular interest as MoARD's 
chief dissemination technology, the catch- 
ment area approach, was the only one used 
in both Luo and Luhya areas, and suggests 
that the differences we find here may not 
be affected by cultural variables. In Gongo, 
poor men ranked MoARD as high as 75 per- 
cent. However, their qualitative assessments 
of MoARD in the discussions were not es- 
pecially positive, indicating that although 
MoARD was their preferred organization 
in relative terms, there is still considerable 
room for improvement. In contrast, non- 
poor women like it. They said, unlike other 
groups, that farmers were chosen equitably 
for assistance. 

Differentiation of Dissemination 
Effectiveness According to 
Social Groups: Evidence 
from Quantitative Surveys 

Cross-tabs were run to analyze whether the 
poor were as likely to receive information 
as the less poor from the different sources 
of information. Recall that three wealth 
categories were formed: (1) the ability of 
assets to meet basic needs (asset coverage), 
(2) index of household relative wealth levels, 
and (3) enumerator evaluation. The first two 
measures were converted into categorical 
variables with three wealth levels. For all 
three measures, the number of households 
in the highest wealth category is less than in 
the other categories (about 10-15 percent of 
all households). We show the results for all 
three wealth variables because they do dif- 
fer somewhat in the way that they differen- 
tiate households (see Table 8.4). 

In general, there are few statistically 
significant results. Farmer-to-farmer contact 
does not appear to discriminate at all across 
different wealth groups. Extension was 
similarly found not to favor the wealthy 
farmers, contrary to widely perceived be- 
liefs. In fact, according to one wealth mea- 
sure (household perception), extension actu- 
ally favored the poorer groups in terms of 

direct contact. This lack of bias in govern- 
ment extension toward the wealthy sharply 
differs from the region's past. Between 
1960 and the late 1970s, Kenyan develop- 
ment programs during the colonial and 
post-independence period favored wealth- 
ier farmers (Leonard 1991). This change is 
likely due to the current greater concern 
with poverty that induced both the govern- 
ment and donors to emphasize more farmer 
participation in extension. While our quali- 
tative research also showed a lack of bias in 
extension toward wealthier farmers, it also 
found that poor and less poor farmers had 
different perceptions as to how they were 
affected by dissemination methods. For 
example, poor farmers felt they were dis- 
advantaged in some aspects of group-based 
methods (see later). 

For CBOs, NGOs, and ICRAF, there is 
some evidence that the more wealthy house- 
holds benefited from direct contact for SFR 
information. For CBOs and NGOs, a posi- 
tive and significant relationship existed for 
one of the three wealth indicators only 
(asset coverage for CBOs and enumerator 
evaluation for NGOs). At the same time, 
other wealth indicators showed evidence of 
the contrary. Thus, the data are not clear for 
these two types of organizations. The evi- 
dence seems strongest in the case of ICRAF, 
where a positive relationship between con- 
tact and wealth level is found for two of 
the three wealth indicators. For example, 
ICRAF reached between 42 and 47 percent 
of the wealthiest group of farmers but only 
about 23-25 percent of the poorest. It is not 
surprising that ICRAF tended to favor the 
wealthy more than others in these non-pilot 
villages in that their familiarity with the 
local population is the weakest of the differ- 
ent sources and the time that ICRAF could 
put into developing a village profile is 
highly limited. 

Cross-tabs were also run to identify any 
patterns between awareness, information 
contact, and gender. No significant rela- 
tionships were found between gender and 
information flow for the entire sample or by 


Table 8.4 Percentage of households with direct contact for SFR 
information, by source and wealth group 

Wealth level 

Source and wealth indicator 





Asset coverage 




Household measure 3 




Enumerator evaluation" 





Asset coverage 




Household measure 




Enumerator evaluation" 





Asset coverage" 




Household measure 




Enumerator evaluation 





Asset coverage 




Household measure" 




Enumerator evaluation 




Other farmers 

Asset coverage 




Household measure 




Enumerator evaluation 




"Significant at .05 level or lower. 

individual source. Thus, women do not ap- 
pear to be discriminated against in receiving 
information. The reason may well be that 
women are, by and large, the major farmers 
in the study sites and men may often be 
absent from the farms. 

Perceived Problems 
with the Approaches 

Perhaps the main criticism of outside dis- 
seminators raised by participants in the focus 
groups is that they leave too soon: "What 
limits full implementation is that they are 
usually left before standing on their feet" 
(NPW, Gongo). Among all the organizations, 
only ICRAF is not mentioned as leaving 
too early, though in the case studies, ICRAF 
received the same criticism. In the focus 
groups as well, some residents reported that 

ICRAF, like the others, did not follow up 
with communities as they should have. 
MoARD, KARI, CARE, and KWAP were 
mentioned as leaving too early by at least 
one of the four focus group divisions (poor 
and non-poor, men and women) in each vil- 
lage. One of the consequences is that groups 
founded by these disseminators collapsed 
once the project ends, and residents feel 

This is compounded because people feel 
disseminators do not allocate enough staff 
to their projects: "Since the extension staffs 
were few, the only staff allocated to the di- 
vision could not reach out to every farmer" 
(PM, Sauri). Residents criticized all outside 
disseminators at some point for this. All of 
these methods emphasized smaller extension 
staffs and more use of local groups. In the 


case of Gongo, for example, MoARD 
founded a number of groups, but lacked the 
trained staff to answer questions. In con- 
trast, the one organization in Gongo that 
provided the most extension staff, the CDC 
(see case study in Appendix B) was by far 
the most popular. 

Teaching and Dissemination 
Methods Employed by 
External Organizations 

The last section introduced some aspects 
of dissemination methods because residents 
often do not separate their assessments of 
outside disseminators from the methods they 
employ. In the last section, however, we 
attempted to evaluate the popularity of 
disseminating organizations on the basis of 
their overall methodological package. It is 
also important to assess the specific methods 
outside organizations use and their popular- 
ity in these communities. These reveal an- 
other complex pattern of relationships that 
is again associated with gender and class, 
and to a lesser extent ethnicity. Specifically, 
people, and especially poor people, prefer a 
mixed approach of formal and informal 
methods. The following material draws on 
the focus group results. 

Methodologies of the 
Disseminating Organisms 

At first glance, the five approaches employed 
by disseminating organizations appear 
complex. Each of them has an extensive and 
well thought out plan for the communities. 
But these methods share a number of fea- 
tures and can be broken down into a number 
of small components that are responsible for 
their relative successes. As a result, they 
also share a number of problems that can be 

Each of the dissemination approaches 
seeks to spread technologies by a combina- 
tion of methods. Broadly speaking these all 
seek to correct widespread criticisms of 
earlier dissemination approaches that were 
found to be overly technical, to not listen to 

local people, to be insufficiently aware of 
local ecological and land tenure conditions, 
and to favor better-off farmers. Newer ap- 
proaches sought to correct this by seeking 
to talk to farmers about their realities, deter- 
mining whether the technologies were use- 
ful, and learning about farmers' conditions 
and needs. This was done through working 
with individual farmers and local groups 
that were to take on the work of adaptation 
and dissemination. 

The methods used in western Kenya are 
described below. The common element is 
that all involve interactions between farm- 
ers and resource persons, the latter of whom 
are often researchers, staff of NGOs, or 

Demonstrations. These are training ses- 
sions designed to convey information about 
a specific topic and it is usually done by dis- 
playing how a technique works or through 
encouraging the farmers (trainees) to try out 
the skills themselves. The aim of this is to 
impart the skills to them. The skills taught 
in the demonstration could be very narrow, 
such as pruning of branches, or more gen- 
eral, such as management of pests. They 
could be conducted at a research station, 
but in western Kenya it is more likely that 
they are conducted in the villages, either at 
a researcher-managed plot or on a farmer's 

Field Days. These are organized sessions 
where farmers meet resource persons, be 
they extensionists or researchers, to discuss 
important topics. Field days often involve 
both discussions and demonstrations on a 
set of technologies. They may be conducted 
at the farmers' fields, or at research or ex- 
tension demonstration plots. 

Farmer Tours. These are planned visits 
to one or more farms, usually by a group of 
other farmers. The idea here is for a farmer 
to demonstrate a farming technique and to 
train other farmers. The fact that such a 
practice is being undertaken successfully 


by another farmer usually instills confidence 
in the trainees that they can also do the same 
on their farms. The visits are usually to 
farms where the farming practice/technique 
being promoted is working well. Several 
farmers, stations, or organizations may be 
visited at a time and the tour itself may last 
for longer than a day. 

Farms/Farmer Visits. These are organized 
routine visits to see the farmers/farms where 
researchers or extensionists or development 
workers are undertaking some activities 
with the farmers for the purpose of monitor- 
ing and evaluating the performance of the 
farming practices/techniques that is being 

Barazas. These are "official" meetings for 
the community called by a chief (govern- 
ment appointed administrator of a location) 
or a subchief (government appointed ad- 
ministrator of a sub-location) or any other 
provincial administrator. Some leaders call 
regular barazas that may cover a range of 
topics. Others may call special barazas to 
handle specific issues. New information 
about agriculture could be the subject of a 
baraza. In most cases the administrator calls 
the officials from different government de- 
partments and occasionally other develop- 
ment workers are also called to discuss im- 
portant topics that are seen to be relevant 
and crucial to the community at that partic- 
ular time. In most barazas, "the community 
is addressed by the experts." 

Meetings. These can be formal or infor- 
mal forums for interactions between different 
parties. They can be held in a meeting room 
or just under a tree and can also target spe- 
cific individuals or be more open. For most 
of the meetings, one person is appointed 
to coordinate and moderate the discussions. 
Meetings are held to discuss upcoming plans, 
to resolve issues, or to monitor progress and 
identify needs. 

Observations. While oral and written com- 
munication methods are important, first- 

hand observation is also key for learning 
to take place. Dissemination methods, such 
as farmer visits, tours, and demonstrations, 
are all designed so that farmers can observe 
how certain practices are done so that they 
may be more easily replicated on their own 
farms. Observation also takes place infor- 
mally, where farmers observe the farms of 
their friends, family, and neighbors, or where 
farmers observe the fields of the "contact 
farmer," an individual experimenting under 
the supervision of an external disseminator 
or of members of committees of SFR -related 
local groups. 

Oral Conversation. Oral conversation is 
the most common vehicle through which 
information is shared because it is the most 
flexible and responsive vehicle in which to 
handle a variety of information needs. Field 
technicians employed this daily to field ques- 
tions from farmers and try to provide an- 
swers related to the management of agro- 
forestry systems. Oral conversation is also 
the main communication method used in 
all other dissemination approaches, such as 
meetings, field visits, and so forth. Oral con- 
versation also takes place informally among 
family members, neighbors, and friends. 

Training. This is a general term that covers 
all aspects of training from group training to 
individual training. It occurs at all levels, 
since researchers, NGO staff, extension, 
and farmers have all received some form of 
training on agroforestry. Also covered under 
training is the production of training ma- 
terials, such as extension guides. Training 
can involve other discussed dissemination 
methods, such as demonstrations tours and 
field days. 

Exchange Visits. This is similar to a 
farmer tour, but one in which reciprocal 
visits are arranged. The most common one 
is where a group of farmers from Village A 
visits farmers in Village B and later on the 
farmers from Village B will visit those in 
Village A. These can take place across in- 
ternational borders and staff from research 


centers or NGOs often accompanies the 

Evaluation of Teaching Methods 
in the Focus Groups 

In the analysis of the PRA exercises that 
follow, the methods used can be categorized 
in a three-part typology of meetings, infor- 
mal methods, and interactive extension 
methods. A special category of meetings 
is barazas. In all communities, external 
organizations entered through contacts with 
chiefs who called people to barazas, which 
are open to everyone in the community. In- 
formal methods represent the oldest of the 
three categories. Farmers frequently spoke 
of learning about new technologies in one 
of two ways. They would observe their 
neighbors and copy them if they liked what 
they saw. They would also talk to their 
neighbors, and adopt technologies if the 
reports were positive. The third category, 
interactive extension, involves some tradi- 
tional ways that outside organizations dis- 
seminate technology, although the way they 
are used in western Kenya often involves a 
more innovative dimension than the older 
conventional approaches. In these commu- 
nities, farmers spoke of the field days, 
demonstrations, tours of model farms, and 
exchange visits with other communities as 
means to learn about new technologies. 
Residents also listed two formal extension 
methods: farmer visits by extension staff, 
and hands-on training of farmers by exten- 
sion staff. These, too, are very popular in the 
studied communities, where some partici- 
pants asked that more training be done 
through these individual farmer visits. 

For the most part, the picture painted 
is one of information flowing mainly from 
disseminator to farmer, and less in the other 
direction. Most of the organizational maps 
showed one-way rather than two-way ar- 
rows. Even in Bukhalalire, where dissem- 
ination methods were most popular, most 
focus groups said that information mainly 
flowed from disseminator to farmer. How- 
ever, some degree of farmer input was so- 

licited in all approaches, and groups from 
at least three of the six villages mentioned 
this specifically: in Sauri, some participants 
reported that ICRAF allowed them to help 
develop the approach, and both poor men 
and women noted that ICRAF asked for 
their input in developing training manuals. 
In Mutsulio, disseminators (MoARD and 
KARI) were said to have asked for farmers' 
ideas and suggestions: "in the initial stages, 
when contact is strong, our ideas are usually 
taken into consideration" (PW, Mutsulio). 
In Mwitubi, contact farmers helped in de- 
veloping training materials. It is likely that 
where contact farmers are used, they had 
input into the technology, although most 
participants in the focus group were not 
familiar with this process. 

Taking the average across the 24 focus 
groups, the overall breakdown of prefer- 
ences for dissemination methods is shown 
in Figure 8.4. People's perceptions of each 
of these methods are described below. 

Meetings. Disseminating organizations 
normally enter the villages via the assistant 
chief, who holds a baraza, or mass meeting, 
open to all. Barazas are thus seen by resi- 
dents as a very important method of dis- 
semination, as it is the one from which they 
are most likely to be included. Slightly less 
popular in these communities are "general 
meetings," which are similar to barazas, 
except that they are called specifically by 
the outside organizations rather than the 
assistant chief. Disseminating organizations 
also encounter a range of local organiza- 
tions, from church associations to women's 
groups to funeral associations, which they 
use as a means of disseminating technolo- 
gies. Other organizations initiate groups for 
this purpose. Taken as a whole, meetings 
(be they barazas, general, or group) are a 
very common place for residents to receive 
information. Among the groups that list 
each method, meetings are ranked at 20 per- 
cent in terms of their usefulness and effec- 
tiveness; barazas, at 22 percent; and general 
meetings, at 19 percent. This is considered 

100 CHAPTER 8 

Figure 8.4 Average preferences for dissemination methods used by external 
organizations in six villages 




Exchange \ 


isits 9% 

Field days 


^^^\ Farmer 











high, given that each of the formal methods 
receives a score between 15 and 5 percent. 
However, these scores are strongly influ- 
enced by two communities. Mutsulio ranked 
group meetings at 55 percent, or higher than 
all the other communities combined, and 
the median score for group meetings was 
about 15 percent. Rankings did not vary 
systematically by ethnicity for any of these 
three methods. 

Rankings vary considerably owing to 
gender and income. Perhaps the most notice- 
able difference to emerge is that women 
score group meetings somewhat higher 
than men. However, case studies and focus 
group discussions indicate that women also 
feel marginalized in group meetings where 
men are present. It may be that in the 
scoring, women were considering women's 
group meetings. This difference, however, 
is most greatly influenced by poor men, 
who rank group meetings at 6 percent. This 
may represent the same point made earlier, 
that poor men may feel more excluded 
from group meetings due to the role of elite 
men. A similar difference, however, has a 
less clear explanation. Women rank barazas 
higher than men, at approximately 27 per- 

cent compared to 16 percent for men. No 
clear differences existed when these rank- 
ings were further broken down by non-poor 
and poor. 

Informal Methods. Also highly important 
for dissemination of information from ex- 
ternal organizations are methods that they 
do not have to plan. Non-poor women in 
Gongo said that "few people learn from for- 
mal ways . . . but many do so informally 
through observation on other farmers' farms 
or orally from other farmers." Among groups 
that included these methods in their scoring, 
each method received approximately 20 per- 
cent. As with other methods, these assess- 
ments vary by village, but less by ethnicity, 
gender, or wealth. These findings on the im- 
portance of informal methods support other 
studies that have found social networks to 
be very important to the diffusion of other 
types of innovations in the region. For exam- 
ple, informal women's groups were found to 
have facilitated adoption of birth control 
where cultural values and beliefs discour- 
aged adoption, and where other programs 
had failed (Rodgers et al. 2001; Behrman, 
Kohler, and Cotts Watkins 2002). 35 

35 Networks succeeded despite cultural values favoring large families, beliefs that oral contraceptives decrease 
women's physical beauty, and rules that prevent women from visiting outside their households. They even suc- 
ceeded in spite of the fact that most women's networks that discussed birth control were very small, composed of 
three or four friends (Rodgers et al. 2001; Behrman, Kohler, and Cotts Watkins 2002). 


The main way that farmers' assessments 
of these informal methods vary is by loca- 
tion, where informal methods are more 
important in Luhya communities. Mutsulio 
residents rank conversation at 41 percent. 
This is particularly important for men, 
who rank conversation at over 50 percent. 
Similarly, conversation is very popular in 
Mwitubi, although not mentioned at all in 
Bukhalalire. In contrast, no Luo community 
scores this higher than 15 percent. For the 
informal method observation, two Luo vil- 
lages, Arude and Gongo, rank this at 21 and 
22 percent, respectively. This might indicate 
a Luo preference for observation, except 
that Sauri ranks observation at 4 percent. No 
Luhya community ranks observation higher 
than 10 percent. As noted elsewhere in this 
study, it is difficult to conclude that there 
are ethnicity-based differences, though this 
would be worthy of further exploration. 

No clear, significant pattern emerges 
from looking at different focus groups' as- 
sessments when gender and wealth are 
considered together across the communities. 
The possible exceptions to this are that non- 
poor men prefer conversations to observa- 
tion by a score of 19 to 9 percent and poor 
women prefer observations to conversations 
by a score of 21 to 13 percent. This may 
suggest that non-poor men are most able 
to rely on their social ties for information 
of this nature. It also may suggest that poor 
women are more comfortable hanging back 
and viewing others before trying something 
themselves. Possibly this is because they 
have fewer social ties that they can use, and 
possibly it is because they are more reluc- 
tant to take risks, as has been established 
to be the case for poor farmers with greater 
degrees of vulnerability in many contexts. 
There may also be a factor of education and 
experience at play, where wealthier farmers 
may be more easily able to digest verbal 
descriptions and convert them to feasible 
plans. These figures are also consistent with 
the general assessments of observations 
by gender. Women rank them at 19 percent. 
Men rank them at 10 percent. However, it is 

not entirely clear that these methods occur 
separately, that is, people probably observe 
and converse at the same time. Nevertheless, 
people listed these as two different ways of 
receiving information, and gave them dif- 
ferent scores in many cases. 

Interactive Extension Methods. Among 
those focus groups that mentioned them, 
farmers claimed that, taken together, demon- 
strations, visits by extension staff, tours, 
field days, and exchange visits account for 
about two thirds of the useful information 
they obtain from outside sources. This is of 
interest because some of these are formal 
agricultural extension methods that are 
criticized in much development literature, 
although in western Kenya many of these 
methods have evolved to include innovative 
dimensions. Some of the weight given to 
these methods is attributable to the fact that 
there are more of them and they are men- 
tioned in more villages. Nonetheless, these 
scores are important because they indicate 
that people like these methods, which is 
supported by the positive comments made 
in the focus groups. People found them es- 
pecially likable as they allowed farmers to 
have direct contact with disseminators, and 
avoided the problems associated with groups 
and meetings. This reinforces the key chal- 
lenge for dissemination — how to balance 
the need for engagement with individual 
farmers with the need to reach a large 
number of them. 

No clear patterns exist in farmer prefer- 
ences for these extension methods. They are 
somewhat more popular in the Luhya vil- 
lages where, on average, they collectively 
score 10 points higher at 60 percent. The 
only notable difference was field days, which 
the Luhya villages collectively scored at 
22 percent versus the Luo villages at 9 per- 
cent. But none of the other possible patterns 
of subgroups vary by more than 5 percent. 
Generally all farmers rate these methods at 
15 percent. The exception is exchange 
visits, which residents hold responsible for 
6 percent of their information, probably 

102 CHAPTER 8 

suggesting that these do not occur often, 
rather than that they are not useful. 

Ultimately, the most important finding to 
emerge from focus groups' discussions of 
these formal methods is that they are impor- 
tant and that local people value them a great 
deal. In some focus groups, people specifi- 
cally said that they would prefer more visits 
in their homes, rather than group meetings, 
for example, non-poor women in Muhanda 
who said that "Meetings are too far away 
and people conduct business there instead 
of train. ICRAF staffs should walk from 
home to home, talk to people and hear their 
suggestions." Mutsulio also offers an inter- 
esting case: there, the PLAR approach was 
used that uses the least amount of traditional 
dissemination methods and places its re- 
liance on local groups. Even there, formal 
extension methods were scored at 45 per- 
cent in terms of importance as a source of 

Methods and Outside Organizations. The 
six communities in this study present a wide 
range of dissemination approaches, some of 
which are quite different. Nonetheless, all of 
them use a combination of group presenta- 
tions, informal methods, and formal exten- 
sion approaches to disseminate information. 

Furthermore, residents seem to like the com- 
bination. Rather than try to emphasize some 
at the expense of the others, for example, 
working through small groups rather than 
farmer visits or mass meetings, dissemi- 
nating organizations will probably be most 
successful when they use this diversity of 

Some logistical problems were raised 
that can be easily solved. These included 
meeting times interfering with funeral and 
market days, dissemination staff arriving 
late, overly long meetings, and use of 
Swahili. In Sauri and Mutsulio, all four 
focus groups complained that participating 
in groups interfered with attendance at fu- 
nerals or church services, since their meet- 
ing times frequently overlapped. Non-poor 
men in Sauri stated that, "lateness in atten- 
dance to barazas reduced the hours that 
farmers would have otherwise learnt while 
at the barazas," as well as a number of other 
scheduling problems that may affect this 
method elsewhere. In Mutsulio, poor women 
noted that they interfered with domestic 
chores as the meetings could stretch for four 
hours. In Sauri, both non-poor groups com- 
mented that the use of Swahili, rather than 
local languages, limits understanding. 


Human and Social Capital Formation: 
Dissemination within the Villages 

In each of the dissemination approaches, the external organization introduces technologies 
and conducts training. However, these organizations cannot reach all farmers effectively, 
and a range of local institutions can be used to further the process. Furthermore, one 
objective of these approaches is to build capacity within the villages, including human and 
social capital, so that residents can continue carrying out dissemination activities with other 
farmers, and eventually in other villages. This chapter looks at farmers' perceptions of the 
various internal methods of dissemination. It then examines the experience of the local groups 
intended to disseminate SFR, where these have been successful, and the types of problems 
they face. Two other internal methods of dissemination are then looked at more closely: the 
"contact farmer" method, and teaching through the schools. Finally, the chapter examines 
knowledge gained about SFR technologies — ultimately one of the most important measures 
of success in dissemination. 

Forms of Dissemination in the Village: Local Groups, 
Contact Farmers, Schools, and Other Methods 

There are several means by which dissemination takes place using local institutions. These 
were referred to in the focus groups as "internal providers." The sources mentioned in all 
six villages were barazas, other farmers, and schools, with some form of SFR group created 
or utilized by outside organizations mentioned in five. Local leaders were mentioned in 
four villages, and women's groups in three. The contact farmer was mentioned as a source of 
information in PRA exercises in only two villages, seen as useful in relative terms in Mwitubi, 
and not so in Muhanda-Arude. Funerals were said to be a source of information in Gongo, 
Mwitubi, and Bukhalalire. 

Barazas are meetings called by the chief and open to all villagers. These are used by the 
external organizations for meetings and training activities. They are also used within the vil- 
lage for ongoing activities, and focus group participants always listed them as "internal 
providers." Other farmers are an informal source of learning, meaning that farmers teach each 
other either actively or passively as farmers observe each other's fields. Schools refer to 
programs that external organizations have normally initiated, where teachers or outsiders con- 
duct training within the schools. Most commonly, the training is given to schoolchildren, who, 
in turn, are expected to do planting at home and to teach their parents. "Local leaders" refers 
to administrators, chiefs, and other leaders recognized within a given village. The contact 
farmer, also referred to in some interventions as an adaptive research farmer, is an individual 
chosen either by the external organization or by community members to use the technology 


104 CHAPTER 9 

on his or her farm, with considerable as- 
sistance given directly to that farmer. This 
serves two main purposes: (1) the contact 
farmer experiments with using the technol- 
ogy under local biophysical conditions, dis- 
covering what works and what does not and 
adapting the technology along the way; and 
(2) other farmers learn from and are assisted 
by the contact farmer. 

"SFR group" is not a term used by the 
disseminators; rather, we use it to refer to 
local groups that take on disseminating ac- 
tivities: actual names include village com- 
mittees, catchment committees, umbrella 
groups, and others, and sometimes villagers 
just refer to them as groups or committees. 
They take different forms and sometimes 
overlap with other internal provider cate- 
gories, for example, women's groups — either 
preexisting or newly formed through the 
intervention — may become the SFR group 
for women. Because of the importance of 
the groups within the different dissemina- 
tion approaches, and the implications they 
have for human and social capital, consider- 
able attention to the experience and evalua- 
tions of these groups is given. 

Quantitative Assessments 
from the PRA Exercises 

Aggregating all villages, there is not a great 
deal of variation in ratings of the importance/ 
usefulness of these internal providers, rang- 
ing from 9 percent to 17 percent of total 
preferences. Contact farmers, barazas, SFR 
groups, local leaders, and other farmers 
were given higher ratings than church and 
women's groups and schools. Barazas rank 
the most consistently high across the vil- 
lages. Only the two Luhya villages of 
Mwitubi and Mutsulio saw barazas as less 
important than at least one other internal 
dissemination method. On average across the 
six villages, there was no difference between 
non-poor and poor farmers, and barazas 
were only slightly more favored by men 
(highest rating) than women (second high- 
est). The popularity of barazas can be under- 

stood in terms of their being community- 
wide meetings from which no one is ex- 
cluded, and a forum that people are accus- 
tomed to attending. However, although 
residents score this method high and it is 
very good for imparting information, it is not 
good for exchange of information among 
participants nor is it particularly regular 
enough to really demonstrate and follow up 
with support for a new innovation. 

The SFR groups were generally viewed 
as important sources of information across 
the villages, especially in Mutsulio, which 
reported groups as particularly important 
sources of information across all four focus 
groups. This is in spite of many problems 
reported with the groups (see later). The 
exceptions were Arude, which ranked SFR 
groups low, and Mwitubi, which did not 
mention them in the ranking exercise, al- 
though they did mention them in the discus- 
sion. Arude was the one village that ranked 
women's groups as important, however, 
indicating that the women's groups worked 
better than those for men. A more detailed 
discussion of findings on the SFR groups 
is found later in this chapter. 

The only other source that Arude ranked 
highly was "other farmers." "Other farmers" 
implies an informal approach to learning 
where people learn from others in informal 
settings through observation or conversation. 
Non-poor men in Sauri also gave a very high 
mark to "other farmers" and to the schools. 
Non-poor men in Mwitubi had a particu- 
larly good impression of the church group. 
These last two findings probably reflect 
particularities of the villages rather than an 
explainable pattern (although they are con- 
sistent with the earlier findings on external 

Contact farmers were mentioned in the 
scoring exercise in three of the six commu- 
nities, and only by poor men in Arude, by 
the women's focus groups in Mwitubi and 
Mutsulio, with low scores in the latter. Con- 
tact farmers are mainly mentioned as the 
entry point of contact for outside organiza- 


Figure 9.1 Average assessments of relative importance of internal disseminators in 
poor versus non-poor focus groups 




I I Poor 

| | Non-poor 

Barazas Other Schools Contact Women's Church Local Children SFR 
farmers farmer group leaders, groups 


tions into the village. They are not credited 
with disseminating much information to 
other farmers. Three villages revealed spe- 
cific cases of dissatisfaction with the contact 
farmer (in one of these, only the poor men's 
group), while the other three were positive 
or neutral. They appear in the organizational 
maps in Arude, Mutsulio, and Mwitubi, 
having links with either other farmers or 
groups in the latter two villages. Insight pro- 
vided by the focus group discussions into 
people's perceptions of contact farmers, in- 
cluding their usefulness and tensions they 
introduce, are discussed later in this chapter. 

Ethnic Group Differences. It is difficult to 
see significant differences emerging between 
the Luo and Luhya villages. Aggregating 
the villages by ethnic groups, there was no 
difference in overall evaluations of barazas, 
other farmers, schools, contact farmers, and 
women's groups. The only apparent dif- 
ferences were among women, where Luo 
women had stronger favorable views of 
barazas than Luhya women, though the 
difference was not great, and where some 
Luhya women had a good impression of the 

contact farmer, whereas women in the Luo 
groups did not even recognize them. Simi- 
larly, Luo women did not recognize local 
leaders/elders and children, whereas the 
former were given a high rating by Luhya 
women. However, not too much should be 
read into this, given that only three villages 
per ethnic group are represented. The vil- 
lages were stratified by ethnic group to see 
if any striking differences emerged, but we 
cannot draw conclusions based on the evi- 
dence found. 

Wealth Group Differences. Turning to ag- 
gregations by non-poor and poor farmers, 
there was little difference between these 
groups in evaluating barazas, SFR groups, 
contact farmers, women's groups, and local 
leaders/elders (Figure 9.1). However, non- 
poor farmers evaluated "other farmers," 
schools, and church groups more favorably 
than did poor farmers' groups, on average, 
and poor farmers had a more favorable eval- 
uation of the role of children. Non-poor 
women generally have more favorable 
reviews of the internal disseminators than 
do poor women. That was the case for all 

106 CHAPTER 9 

Figure 9.2 Average assessments of relative importance of internal disseminators in 
men's versus women's focus groups 





I I Average men 
| | Average women 


Barazas Other 














categories except women's groups, church 
groups, and SFR groups (where there were 
equal rankings across wealth status), and 
poor women rated local leaders/elders more 
favorably. This suggests that the group- 
based approaches are working well for poor 
women, where they feel less excluded than 
they do from more conventional approaches 
involving barazas, other farmers, and con- 
tact farmers. Non-poor men tend to favor 
other farmers, schools, women's and church 
groups, and local leaders more than do poor 
men, whereas poor men favored barazas, 
schools, and SFR groups (not greatly) com- 
pared to non-poor men. 

Gender Differences. In the aggregate, 
there were strikingly few differences in the 
evaluations of men and women, suggesting 
that these methods were not discriminating 
based on gender. This is also consistent 
with the survey findings (Figure 9.2). The 
main exception is the contact farmer, where 
women gave a score of 27 compared to 
men's scores of 12. If the contact farmers 
were men in these villages, which is implied 
in the focus group discussions, it may be 
that women did not feel as competitive to- 

ward them, and were less affected by the 
amount of attention they received from out- 
siders, which was one of the two main com- 
plaints against them. It could also mean that 
women did not expect as much from them 
or knew less about their function, so that 
they were less disappointed in the contact 
farmer not sharing information, which was 
the second main complaint against them. 
SFR groups were slightly more favored by 
men than women, which may reflect men's 
greater participation in these groups. The 
picture changes, however, when men and 
women's groups are disaggregated by 
wealth groups. Poor men significantly favor 
barazas, schools, and SFR groups more than 
poor women, whereas poor women prefer 
the contact farmer, women's and church 
groups, and local leaders more than poor 
men. This suggests that gender differences 
are greater among poor farmers than among 
women and men as an undifferentiated 
group. However, there is less difference be- 
tween non-poor men and non-poor women, 
with the only significant difference being 
the contact farmer, which receives a rating 
of 35 from non-poor women but a from 
non-poor men. Church and SFR groups also 


get a somewhat higher rating from non-poor 
men than non-poor women. 

Other Findings Related to Particular Groups 
and Villages. Among the villages, other 
ratings that stand out (40 or over) are in 
Gongo and Bukhalalire, where poor men 
gave notably high ratings to SFR groups as 
compared to the other groups in the village; 
in Mwitubi, where a church group was par- 
ticularly popular among the non-poor men; 
and in Sauri and Arude, where the women's 
groups were successful in the eyes of poor 
women. Non-poor men in Sauri had a very 
favorable impression of the schools program, 
of other farmers, and of the contact farmer. 
Finally, non-poor women in Arude rated 
other farmers with a 70. 

It is also worth pointing out the case of 
SFR groups in Mutsulio, where they were 
ranked high by all four groups and given a 
score of 80 by poor women, and in Gongo, 
where they were given ratings of over 40 
by poor men (compared to the average of 
around 10 for the other three groups), and 
Bukhalalire (where the average for the others 
was closer to 20). Notably absent is any 
mention of SFR groups in Mwitubi. 

The Importance and 
Performance of the 
Local SFR Groups 

As discussed earlier, each of the dissemi- 
nation approaches used across these six 
villages relies upon local groups for dissemi- 
nating the technology across a wide group 
of farmers, and for ensuring sustainability. 
For this reason, community perceptions of 
the local SFR groups (hereinafter "groups") 
warrant more detailed discussion. Focus 
group discussions focused heavily on issues 
surrounding groups, in part because of the 
questions asked and in part because par- 
ticipants had strong views on these groups. 
These groups were ranked as a relatively 
important source of information, but they 
have also experienced many problems. 

Access to the Benefits of SRF 
Groups: Group Participation and 
Training by Group Members 

In most cases, the groups were said to have 
provided benefits to their members. How- 
ever, in most villages they appear to be pro- 
viding little information to other farmers. 
One problem is the lack of participation in 
the groups, either because of self-exclusion 
or exclusion by group members. Low levels 
of participation directly in the groups would 
not be as large a problem if the groups were 
conducting dissemination activities with 
other farmers as envisioned. The second 
problem, however, is the lack of training 
and dissemination carried out by the groups 
to other farmers. Five of the six villages 
reported one or both of these problems. 

In Muhanda, both men's focus groups 
said that the local groups introduced social 
tensions, attributed to the fact that some 
village members benefited from the groups 
while others did not. Low levels of partici- 
pation were reported in the groups, and the 
groups were said not to have disseminated 
to other farmers. This is confirmed by the 
absence of any linkages drawn between the 
groups and other farmers in the organiza- 
tional mapping exercise. Non-poor women 
showed a strong two-way information flow 
between a women's group and their rela- 
tives, although neither focus group showed 
a link between the groups and other farmers. 
In the discussions, however, women were 
more positive about the experience with 
groups and their potential for the future. 
Women report having benefited financially 
from the women's groups formed through 
the CARE intervention, although these 
groups did not last after CARE left. Non- 
poor women, however, did say they would 
like to have women's groups dealing with 
agriculture, and that women's church and 
welfare groups also serve as agriculture 
groups in some cases, for example, con- 
tributing maize and beans for funerals. It 
is possible that, especially under conditions 
of an AIDS crisis, groups are strained for 

108 CHAPTER 9 

resources and members strained for time, 
and those with simultaneous welfare func- 
tions, as well as those more established and 
thus more cohesive, may be more enduring. 
In Gongo, a more complicated picture 
emerges from the experience of groups. Here, 
poor farmers said they were less active in 
the groups than non-poor ones, and in the 
focus groups only non-poor men and women 
showed links between groups and farmers. 
At the same time, in the scoring, the groups 
were said to be the most important internal 
source of information for poor men. This 
might imply that poor residents have less 
access to initial introductions of information 
by disseminating organizations that deal with 
groups, and that information filters through 
groups to the poor. In the organizational 
map, poor men showed MoARD as respon- 
sible for many types of training, but not 
with direct links to groups or farmers; poor 
women did show a weak two-way link be- 
tween MoARD and groups. Non-poor men 
and women drew strong two-way inter- 
actions between MoARD and both internal 
groups and community members. In Sauri, 
residents scored group meetings as compar- 
atively less important as sources of informa- 
tion. In the discussions, rather than saying 
that the groups did dissemination, the men's 
focus groups mentioned that members "lead 
by example." This implies that group mem- 
bers adopt technologies and that others are 
intended to observe them. This is one of the 
intended means of dissemination, but group 
dissemination activities are intended to go 
beyond this. It is important, however, to note 
that three of the four groups' organizational 
maps (the exception was poor men) showed 
strong two-way links between groups and 
other farmers, implying that dissemination 
was taking place. Additional positive out- 
comes of group activities did emerge, how- 
ever, which are reported later in the section 
on social relationships. An additional issue 
emerged in Sauri concerning men's domi- 
nance and its effect on women's participa- 
tion in groups. In the focus groups, poor and 
non-poor women said they felt intimidated 

by men at groups, and that this limited their 
ability to learn from them. This reiterates 
the importance of having separate groups 
for men and women. 

The issues emerging from comparing 
Luo and Luhya villages were similar in both 
positive and negative findings. This indi- 
cates that groups were not necessarily a 
more effective means of disseminating lo- 
cally within one ethnic group more than an- 
other, as was initially thought might be the 
case. For instance, in Mutsulio, a Luhya vil- 
lage, the groups formed through the PLAR 
methodology are visible and ranked as an 
important source of information compared 
to other villages, especially by poor women 
who gave them a score of 70, and who com- 
mented that "committee members partici- 
pated very much in organizing and mobiliz- 
ing farmers." In the organizational maps, it 
is also the women that showed links be- 
tween groups and farmers. Three of the four 
focus groups said the groups were effective, 
with one saying that the groups were con- 
tinuing with their work after the external or- 
ganization left, and motivated to work 
harder as yields increased. Nevertheless, in 
the discussions, group members were said 
to have disproportionately benefited from 
the process. Most other farmers in the com- 
munities were said not to participate in the 
groups, and those that do are the better off. 
Poor men note that they do not have enough 
land to become group members. As in 
Sauri, observation of group members' fields 
was the main way in which the groups ap- 
proached dissemination: members were 
said to "envision commitment and hard 
work as ways to spread technology, so that 
other farmers can observe the technologies 
as practiced by the committee members." 
Three of the four Mutsulio focus groups re- 
ported that the committees did not talk to 
farmers as much as they said they would, 
and often ignored their requests. Interest- 
ingly, while women rank groups highly as a 
source of information, they also say that the 
groups are forums for men or are dominated 
by men. Non-poor women expressed reluc- 


tance to participate for this reason, again af- 
firming the value of separate groups for 
men and women. 36 

In Mwitubi, women had more positive 
evaluations of the groups, whereas men 
tended to have a more negative view. In 
the organizational maps, three of the four 
groups showed links between the groups 
and farmers, although oddly only the men's 
group mentioned women's groups (poor 
women emphasized church groups) and only 
poor women mentioned the catchment com- 
mittee. In the discussions, however, all four 
focus groups were familiar with the village 
committees, and there were no indications of 
greater or lesser access to the committees, 
with the exception that non-poor women 
thought that farmers with more money were 
the ones who participated in committees 
and groups. However, both men's focus 
groups said that the committee did not teach 
or advise other farmers, although poor 
men said that village committee members 
had trained their children, neighbors, and 
friends. Women's views were different. 
Non-poor women indicated that the com- 
mittee catered to farmers' needs, and that 
members reached out to nonmembers with 
information and the latter were eager and 
willing to learn. Poor women indicated that 
non-committee members gain skills by ob- 
serving trials on members' farms. Women 
saw new SFR groups, in particular women's 
groups, as critical to extension and argued 
that some have taken up that role and were 
disseminating with the support of the chief 
and his assistant. Women come across as 
pragmatic in their views of and contribu- 
tions to extension work in the community. It 
is women's groups that turn out to be con- 
tinuing with dissemination work and not the 
village committee, although the latter had 
been formed for the sole purpose of dissem- 
ination. This suggests a finding similar to 

that in Muhanda, that existing groups seem 
to work better than new ones formed for this 
intervention. With regard to groups, there 
were no major differences of opinion be- 
tween poor and non-poor participants of 
either gender. Bukhalalire presents a very 
different picture from the other villages. All 
maps show the umbrella groups, and though 
more links are shown with schools than 
directly with farmers, the discussions reveal 
considerable success with the use of groups. 
Moreover, it is the only one of the six vil- 
lages that reported that groups were involved 
with dissemination outside the village as 
well as inside, which is the long-term goal 
of ICRAF and its partners in the SFR inter- 
ventions. It is not possible to know the ex- 
tent to which this success is a result of the 
umbrella group approach, or a more socially 
cohesive community, although probably both 
factors play a role. According to participants, 
preexisting groups were involved and new 
women's groups were formed through the 
intervention, and across villages the groups 
elected representatives to the umbrella 
group committee that coordinates activities 
throughout the catchment. There does not 
appear to be visible gender- and wealth- 
related differences. Benefits from the group 
were said not to depend on one's social or 
economic status and benefits were said to 
have been equally accessible. All four focus 
groups described training and dissemina- 
tion that the groups were carrying out, in- 
side and outside the catchment, to other 
groups and schools. Women were particu- 
larly descriptive and vocal about the group's 
activities. Non-poor women said that the 
umbrella group is useful in dissemination 
through demonstrations and field days, and 
that neighboring villages learn from them 
informally and through attendance in dissem- 
ination sessions. Poor and non-poor women 
said that nonmembers of the umbrella groups 

36 This finding on the effectiveness of women's groups for reaching women is supported in the studies referred to 
in Chapter 8 (Rodgers et al. 2001 ; Behrman, Kohler, and Cotts Watkins 2002) on the importance of women's in- 
formal networks. 

110 CHAPTER 9 

attend and learn in the field days, which they 
organize without staff of external organiza- 
tions. Participants were also of the view that 
farmers had been involved in developing 
training materials and that the groups had 
continued with dissemination work. 

Social Capital, Social 
Relationships, and Power 

One of the study hypotheses 37 was that dis- 
semination through local groups will en- 
hance a community's social capital — the 
social networks, relationships, and organi- 
zations that facilitate access to resources, 
provide support, and otherwise enhance the 
well-being of the village and individuals 
who participate in them. Conversely, it is 
possible that the groups introduced new 
social divisions. A second, though weaker 
hypothesis, was that the SFR interventions, 
including the training and group-based, 
might increase the confidence of farmers, 
leading farmers to make more demands of 
the groups and groups to make demands on 
external institutions. At the same time, it 
was hypothesized that the interventions and 
group-based activity could possibly reinforce 
or alter existing power relationships within 
the village. The exclusion of some people 
from the groups as discussed in the previous 
section is one way in which power relation- 
ships were expressed, though these might 
be reassertions of preexisting power relation- 
ships or indications of new ones. Because 
the issues of social relationships, empower- 
ment, and power are also closely related, 
they are addressed together in this section. 

Many of the villages reported that group 
activities related to the SFR interventions had 
increased community solidarity in certain 
respects. In Sauri, all focus groups men- 
tioned that the new groups had brought 
the community closer together in some way. 

However, all focus groups except non-poor 
women implied that local elites were 
stronger as a result of these groups. It may 
be inevitable that to some extent, local elites 
will dominate groups, especially new ones 
formed for managing new resources. At the 
same time, the interventions in Sauri also had 
the effect of increasing farmers' confidence 
— a comment made by all groups except 
non-poor men. According to focus group 
discussions, local people were instrumental 
at changing a number of dissemination 
practices, including "demanding" that local 
committees bring in other technologies such 
as dairy cows, that the area committee send 
members to other villages to learn about 
technology, that ICRAF exert influence over 
local committees when their rules were not 
acceptable, and that meetings be held when 
no funerals or markets were underway. 

In Muhanda, poor men said the exten- 
sion activities had brought the community 
together. They also, however, emphasized 
the importance of making a greater effort 
to bring the entire community together for 
initial introductions to the intervention, so 
that some people do not get excluded (this 
is directed more to the external organiza- 
tions than the groups). However, both men's 
focus groups said that the local groups in- 
troduced social tensions and politics, attrib- 
uted to the fact that some benefited and 
others did not, the extra attention that some 
farmers had from the external organization 
(in particular, the contact farmer), uneven 
distribution of resources, the ability of some 
to amass wealth through the process, and 
conflicts over resources. Women's groups 
formed by the CARE intervention created 
new forms of social capital, but collapsed 
after CARE left, in part because of financial 
mismanagement, which created new social 
tensions among the women affected. 

37 The study did not work with explicit hypotheses; rather it asked a set of research questions. Nevertheless, ques- 
tions about whether social capital is enhanced through the use of groups imply a hypothesis that it might. 


In Gongo, poor and non-poor residents 
had different views of the impact of the 
groups on social relationships. Non-poor 
men and women reported that the presence 
of the groups in the communities strength- 
ened local solidarity. These views must be 
seen within the context of existing power 
structures within the village. Twelve mem- 
bers of the groups formed by MoARD were 
chosen from each of the clans living in 
Gongo. In practice this meant that clan 
elders were selected. Thus, those serving in 
the groups were likely the oldest, wealthiest, 
and most powerful residents. This may ex- 
plain why poor men and women reported 
that local groups were chosen unfairly, and 
asked that this be remedied in the future. 
Non-poor men and women acknowledged 
that group members get benefits first, and 
that they give these benefits to their close 
kin before others. 

In Mutsulio, the dominant picture that 
arose was of better-off farmers as the most 
common group members. Poor men noted 
that they feel a great constraint to partici- 
pating in groups, since they do not have 
enough land to become group members. 
In Mwitubi a mixed picture arises with 
marked gender differences. Non-poor 
women expressed that dissemination had 
brought farmers closer together as a result of 
"sharing in on-farm technology trials" and 
that the technologies "united the farmers." 
Men participants were generally of the 
view that interventions through the groups/ 
committees led to alteration in community 
power relationships. Poor men, in particu- 
lar, pointed to the misunderstandings caused 
by the fact that outside institutions visited 
some farmers often and not other farmers. 
Women participants saw the situation less 
as one where people acquired status through 
the groups, but rather that those who joined 
already had status. This makes sense as the 
propensity to seek community leadership 
positions often hinges on the socioeconomic 
status of an individual. However, poor 
men appear to have acquired some power 

through the process, reporting that farmers 
made demands on the committee when it 
was active and the committee, in turn, made 
demands for extension services on MoARD. 

All four focus groups in Mwitubi men- 
tioned conflict within the groups, particu- 
larly related to rivalry among the leadership. 
Non-poor men added that failure of com- 
mittee members and extension staff to visit 
farmers was another source of conflict. 
However, all the groups appear to agree on 
the fact that the extension interventions led 
to competition and conflict in some ways 
and to cooperation and cohesion in others. 
Overall, the participants seem to have viewed 
the emerging conflicts and competition more 
as learning points than as negative sides of 
extension. One group looked beyond the life 
of the dissemination intervention to argue 
that farmers interact closely even after the 
external disseminating organization had left. 

In considering the effect of the umbrella 
group approach to dissemination on com- 
munity social relations in Bukhalalire, par- 
ticipants across the spectrum agreed that the 
approach had brought community members 
closer together. Poor men said this occurred 
through "discussing and exchanging infor- 
mation about the various technologies" and 
that they "now work mostly as a team." The 
participants also indicated that the groups 
did not favor those of any particular social 
status; rather that benefits of the tech- 
nologies depend ultimately on how hard 
one works. The umbrella group approach 
appears to have supported the evolution of a 
cohesive community that strove to be self- 
reliant through intracommunity exchange of 

The Contact Farmer 

As noted earlier in this chapter, contact 
farmers are mainly seen as the point of con- 
tact for outside organizations, and not much 
more was said about them in most of the 
villages. Focus groups in three of the six 
villages indicated problems with the contact 

112 CHAPTER 9 

farmer method (although in one, only poor 
men were critical). Although most groups 
were positive about the method, they pro- 
vide important insight into unintended social 
consequences that dissemination methods 
can have. First, the contact farmer was seen 
as unfairly receiving too much attention from 
external organizations and thereby gaining 
too much power: "model farmers gained 
more prestige and control over other farm- 
ers as they trained them." Second, contact 
farmers in two villages were not seen to 
have shared information with other farmers. 
In one village, non-poor men said that no 
efforts were made to extend the technology 
to the rest of the catchment members, no 
field day was done, and the rest of the 
catchment members felt excluded. Although 
they could observe what the contact farmer 
was doing, they did not copy and adopt 
what they observed because they were not 
sensitized to believe that what was being 
done was for their own benefit. 

In another village the experience with 
the contact farmer left an unfavorable im- 
pression with three of the four groups (the 
exception was non-poor women). They were 
seen as (1) unfairly favored and (2) keep- 
ing knowledge to themselves rather than 
spreading it among others. Poor and non- 
poor men stressed that the contact or adap- 
tive research farmers do not share informa- 
tion, and according to the better-off men, 
they are seen as "the wealthy and educated 
who are frequently visited and make others 
feel left out and different from the preferred 
farmers." The constant attention from gov- 
ernment and other institutions is said to be a 
problem. There was also the perception that 
the contact farmers were selected by the ex- 
ternal institutions, although according to the 
organizations, they are to be selected by the 
villagers. Poor women criticized the contact 
farmer for not sharing seeds, saying that 
people waste their time trying to get things 
from them. However, both groups of women 
saw one CARE contact farmer in a positive 
light, recalling that he helped with training 
and, more significantly, brought women 

together to form women's groups to pro- 
mote the technology. 

Notwithstanding local perceptions in 
some villages, it is important to recognize 
the role of the contact or adaptive research 
farmer from the perspective of the dissem- 
inating organization. As CARE points out 
in its TRACE approach, technologies that 
work in one region may not be adapted to 
a new region. Adaptive research farmers are 
crucial for testing technologies and practices 
and adapting them to local conditions, be- 
fore they are disseminated to other farmers. 
There may thus be a period in which there is 
considerable contact between the ARF and 
the external organization, before many other 
farmers are brought into the process. It is 
equally important, however, to recognize 
that the social context will affect outcomes, 
and that this method as currently practiced 
is clearly problematic, given local social 
relationships, and has significantly affected 
the way in which people respond to the ex- 
ternal organizations as they introduce tech- 
nologies. It may be necessary then to bring 
the community more widely into the learn- 
ing process at an earlier stage, to make sure 
people understand the role of the ARF and 
approve of the choice. 

Dissemination through 
the Schools 

The schools' approach to dissemination is 
hinged on the premise that students would 
go back home to practice what they learned 
in schools and may influence or train their 
parents and communities to adopt learned 
technologies. Therefore, some of the dis- 
seminating organizations working in west- 
ern Kenya made specific efforts to reach 
schoolchildren with messages on agricul- 
tural technologies. In each of the villages 
under study, there was evidence that school- 
children had been reached with dissemi- 
nation messages by one organization or 
another. The organizations mentioned as 
having reached the schools with dissemi- 
nation include MoARD, KWAP, CARE, 


KARI, Action Aid, and Africa NOW. 
MoARD and CARE reached the schools 
mostly through 4K clubs; KWAP and Action 
Aid Kenya, through the umbrella groups; 
and Africa NOW, directly. Although several 
technologies were disseminated to schools in 
respective villages in a variety of combina- 
tions, tree nursery establishment/management 
and tree planting were the main focus of the 
schools approach across the villages. Only 
in Gongo and Sauri villages were soil and 
water conservation, and potable water har- 
vesting, respectively, disseminated in addi- 
tion to messages on tree growing. 

After learning the various technologies 
in schools, students made efforts to train 
their parents on the same, albeit weakly in 
some instances. For example, the poor men's 
focus group in Gongo village said, 

The extension approaches reach school- 
children from primary to secondary 
schools and had been done normally 
through 4K clubs or by the organiza- 
tions themselves by introducing tree 
nurseries in schools. Students also train 
their parents on what they have learned 
from the schools such as the planting of 
trees, which was introduced by CARE 
to the schools. Parents learn what is 
happening in the schools through their 
children and have not trained anybody 
because schoolchildren do not deeply 
and adequately explain acquired tech- 
nologies to their parents and some par- 
ents do not have children in schools. 

In Muhanda, where CARE also pro- 
moted school programs, the women's focus 
groups said that children came home, made 
vegetable gardens, planted trees at home, 
and trained their parents in kale and tomato 
planting. Children are said to still be prac- 
ticing what they learned. 

Where students have not made specific 
efforts to train their parents, the latter "learn 
through observation in the school com- 
pounds" as the poor men's focus group in 
Bukhalalire village put it. Parents' learning 

by observing the school farms appears to 
be weakening over time because, as the 
non-poor focus discussion group in Gongo 
village said, 

In the past when there was any occasion 
in school, parents were taken round the 
school so that they could see what their 
children planted. These days, parents 
are not learning anything from school 
because even when they attend any 
function in school, they are not taken 
around to see the ongoing projects. 

The poor women's focus group in the 
same village argued that children reach their 
parents with the technologies learned by 
practicing the technologies at home, "al- 
though they hardly convince their parents to 
adopt the technologies." Students had tried 
at home such technologies as kales and 
tomatoes growing in Muhanda and Bukha- 
lalire villages, and had earned income 
from selling their produce to the commu- 
nity. Where the school approach was not in 
place, such as in Mutsulio village, no agri- 
cultural knowledge flowed from the schools 
to the community and students did not prac- 
tice any agricultural activities of their own 
at home. The woodlots established in the 
schools still existed as at the time of this 
study and schools had reaped benefits by 
selling trees in some of the schools. In Sauri 
village, the non-poor men's focused discus- 
sion group argued, "parents only know the 
outcome of sale of trees when called upon 
during parents' meetings and the school 
committee informs them as parents." The 
income earned was used to pay the school 
watchman and buy pieces of chalk. In 
Muhanda, Mwitubi, and Bukhalalire, focus 
groups were agreed that the trees were a 
source of income and building materials for 
the schools. 

The major challenge identified with the 
schools approach was that in some villages, 
students hardly convinced their parents to 
adopt technologies. This is because parents 
looked at the students as children from whom 

114 CHAPTER 9 

no serious idea could originate. In addition, 
it was argued that formal agriculture lessons 
in the schools were theoretical and lacked 
practical exposure, so that students had noth- 
ing tangible to disseminate. Another chal- 
lenge was that livestock from the neighbor- 
hoods of some schools destroyed trees in 
the schools. 

Overall the schools' approach to dissem- 
ination appeared effective in the medium to 
long term as it put agricultural knowledge 
at the fingertips of tomorrow's farmers. This 
way the next generation of farmers would 
be better trained. In the short run, the ap- 
proach has shortcomings in that not all 
parents in the farming community have 
children in school so as to be reached with 
the approach. Moreover, in an African rural 
setting where children are considered igno- 
rant and have no established forums for dis- 
cussion with their parents and other adults 
in the community, the flow of information 
from students to their parents and commu- 
nities is largely hampered. 

Knowledge Acquisition 

Although focus group participants have 
varying opinions of disseminating organiza- 
tions and their methods, the best measure 
for assessing the performance of these or- 
ganizations is the amount of knowledge 
people gained through the dissemination 
efforts. Participants report that they learned 
a great deal about technologies, and are 
generally consistent in this assessment. Like 
all other factors, however, knowledge gain 
varies along with community of residence, 
gender, and wealth. Equally important is that 
people improved themselves personally in a 
variety of ways, owing to their knowledge 

Technologies Introduced 

Villagers learned about many different 
agricultural techniques and technological 
innovations. In Sauri, for example, residents 
learned about schemes to grow and market 

sunflower seeds, maintain rabbit warrens, 
use household rubbish as mulch, and keep 
koi ponds. A more typical example is 
Gongo, where residents learned about 
building latrines, malaria prevention, and 
sanitizing drinking water. All the villages 
in the qualitative dissemination study were 
exposed to a wide range of SFR technolo- 
gies, with a total of nine mentioned across 
them. Of these nine, seven were mentioned 
in all six villages studied, a finding made 
more interesting by the fact that the focus 
group participants themselves generated 
the categories. Everywhere, people men- 
tioned tithonia, farmyard manure, compost, 
commercial fertilizer, rock phosphate, im- 
proved fallow, and terraces as SFR tech- 
nologies. One of the remaining two, crop 
residue, was mentioned in all communities 
except Arude. The final technology, alley 
cropping, is mentioned only in Bukhalalire, 
because it was introduced only there. The 
consistency with which the same tech- 
nologies were mentioned across the villages 
and the high scores for knowledge gained 
indicate strongly that people know about 
and learn about the technologies that orga- 
nizations introduce. 

Knowledge Gained 

As described in the previous chapter, focus 
group participants used "ladders" to indicate 
the amount of knowledge on the technology 
they had before the intervention, the per- 
cent increase after the intervention, and the 
percent increase (or decrease) between the 
end of the intervention and the present. A 
score of means no knowledge and 100 
percent means full knowledge. Figures 9.3, 
9.4, 9.5, and 9.6 show graphic representa- 
tions of the ladders exercise for four focus 
groups. Figures 9.3 and 9.4 compare results 
within Sauri village across poor and non- 
poor women. Figures 9.5 and 9.6 compare 
results within Mutsulio village across poor 
women and poor men. Focus group partici- 
pants already divided by poverty level and 
by men and women were further divided 


Figure 9.3 Sauri poor women's group: At least four years of primary education 

Percentage of knowledge gain 







40 - 



10 - 


Before project 

1 During project 

After project 

C = Compost 

CF = Commercial fertilizers 

CR = Crop residues 

F = Farmyard manure 

IF = Improved fallows 

RP = Rock phosphate 

T = Tithonia 

TE = Terraces 

= No knowledge 

100 = Perception of 

"complete knowledge" 






according to education levels. 38 Most re- 
sults reported here aggregate these groups' 
answers, except where significant differences 
emerged, which are reported in the section 
on "education levels" below. Figure 9.7 
shows average knowledge gain (across the 
four focus groups) in each village for each 
technology. The most surprising finding 
about the amount of knowledge gained is its 

uniformly high level and its consistency. 
Although the range is from 33 percent to a 
massive 76 percent, the last number is for 
alley cropping, a technology that was only 
introduced into one community (Bukha- 
lalire) but appears to have been extremely 
successful in terms of what people learned. 
More typical figures come from other tech- 
nologies. Total knowledge gain on these 

38 In the analysis, no education was combined with up to three years of primary education, considered to be "less 
educated." Those with four years of primary education were combined with those with secondary education, con- 
sidered "more educated." 

116 CHAPTER 9 

Figure 9.4 Sauri non-poor women's group education level: At least four years of primary 

Percentage of knowledge gain 















Before project 

) During project 

After project 

C = Compost 

CF = Commercial fertilizers 

CR = Crop residues 

F = Farmyard manure 

IF = Improved fallows 

RP = Rock phosphate 

T = Tithonia 

TE = Terraces 

= No knowledge 

100 = Perception of 

"complete knowledge" 






varies between 33 percent for rock phos- 
phate and 54 percent for crop residue and 
terraces. Most are clustered around the mean 
of 49 percent knowledge gain. The two low- 
est scoring technologies are rock phosphate 
and commercial fertilizers, but even here 
participants claimed to have increased their 
knowledge by 33 percent and 39 percent, 
respectively. These lower levels of knowl- 
edge acquisition are likely to reflect the fact 
that these technologies are more difficult or 
expensive to obtain, and fewer people have 
much experience using them. 

High levels of knowledge growth were 
achieved in terracing, which in Kenya 
usually involves leaving a vegetative strip 
in place (i.e., not plowing all the land) that 
forms a terrace. Average knowledge gain 
was 53 percent with no village gaining less 
than 43 percent, indicating that those orga- 
nizations that promote it, particularly CARE, 
are doing so to great effect. For the two 
technologies of primary interest to this 
study, tithonia and improved fallow, knowl- 
edge gain was around the average for the 
other technologies, at 51 and 45 percent, 


Figure 9.5 Mutsulio village— poor women: At least four years of primary education 

Percentage of knowledge gain 







40 - 



10 - 

Before project 

I During project 

After project 

C = Compost 

CF = Commercial fertilizers 

CR = Crop residues 

F = Farmyard manure 

IF = Improved fallows 

RP = Rock phosphate 

T = Tithonia 

TE = Terraces 

= No knowledge 

100 = Perception of 

"complete knowledge" 






respectively. Knowledge gains on these 
technologies for all villages are summarized 
in Table 9.1. The highest increase was for 
Bukhalalire, at 61 percent and 69 percent, 
respectively. The average was brought 
down by one village, Gongo, at 14 percent 
and 7 percent, respectively. In Gongo, poor 
and non-poor men's focus groups reported 
problems with the technologies: that titho- 
nia was too labor intensive and improved 
fallows required too much land. Both also 
expressed that the disseminating organi- 
zation did not adequately respond to these 
concerns, and this caused negative evalua- 

tions of the experience and the conclusion 
that little was learned about these tech- 
nologies. In fact, as noted in Chapter 6, very 
few farmers were using the agroforestry 
systems. This points to a relationship be- 
tween good two-way dialogue between 
disseminating organization and farmers, on 
the one hand, and ability to learn about the 
technology, on the other. 

All but 4 of the 24 focus groups reported 
that before the intervention they had no 
knowledge at all of either tithonia or im- 
proved fallows. Although scores of pre- 
intervention 0's are common for other 

118 CHAPTER 9 

Figure 9.6 Mutsulio village— poor men: At least four years of primary education 

Percentage of knowledge gain 































Before project 

t During project 

After project 

C = Compost 

CF = Commercial fertilizers 

F = Farmyard manure 

IF = Improved fallows 

RP = Rock phosphate 

T = Tithonia 

TE = Terraces 

= No knowledge 

100 = Perception of 

"complete knowledge" 


technologies as well, these two stand out as 
being particularly unfamiliar. The fact that 
Bukhalalire claims the highest gains suggests 
that there may be a correlation between 
levels of satisfaction with the dissemination 
process, including success with group- 
based methods, and success in knowledge 

In cases other than Gongo, low rankings 
cannot be readily explained. Crop residue 
was not mentioned in Arude focus groups. 
In addition, rock phosphate scored low in 
Arude (9 percent) and Bukhalalire (22 per- 
cent), but participants did not offer explana- 

tions for this. On the other hand, one com- 
munity stands out as achieving high gains in 
two technologies. Bukhalalire participants 
claimed to gain a 63 percent increase in 
knowledge on farmyard manure and im- 
proved fallow resulting from the interven- 
tion. Similar figures for the average across 
the villages were 46 percent and 45 percent, 
respectively. It was also highest on tithonia 
and improved fallows in Bukhalalire than in 
the other villages. Bukhalalire participants 
generally claim to have learned more than 
the average on most technologies, the ex- 
ception being rock phosphate and commer- 


Figure 9.7 Average knowledge gain for each technology, by village 

Percentage of knowledge gain 


70 " 


50 " 



20 " 

10 " 

| | Average Sauri 

| | Average Gongo 

| | Average Arude 

| | Average Bukhalarire 

I Average Mwitubi 

1 I Average Mutsulio 














cial fertilizer. Given the success with the use 
of groups, there may be a relationship be- 
tween how well the local groups function 
and the success achieved in terms of knowl- 
edge acquisition in the technologies. 

Variations by Education Level of Villagers 
and Age of the Intervention. Two other 
ways in which knowledge gain varies are by 
education level, and by age of the interven- 
tion. Participants with more education gen- 

erally learn more about SFR technologies 
than those with less. Nonetheless, the differ- 
ence is less dramatic than one may expect, 
which indicates that these technologies can 
be used by people regardless of education, 
and that SFR disseminators have achieved 
their goals of reaching some of the more 
vulnerable individuals. 

The PRA exercise reveals that people 
learn between the beginning and end of the 
intervention (knowledge before intervention 

Table 9.1 Knowledge gain for tithonia and improved fallows in the six study villages 
from focus group discussions 

Study village Average knowledge gain for tithonia Average knowledge gain for improved fallows 









120 CHAPTER 9 

compared to knowledge after) than they do 
between the end of the intervention and the 
present. In no cases do people report that 
they learned a greater amount after the in- 
tervention (i.e., through practice) than they 
learned through the intervention (that is, 
through teaching and practice), and in only 
one case, a group reported that they knew 
less now than before the intervention. In al- 
most all cases, people indicate that they con- 
tinued to learn an impressive amount after 
the intervention was over, that is, through 
practice or ongoing dissemination activities. 
This suggests that in spite of the many cri- 
tiques that people have of the dissemination 
processes, organizations, or local groups, 
they are using or at least have learned a 
considerable amount about the technologies, 
and this learning continues after the organi- 
zation departed. In very few cases do people 
report knowing less now than they did after 
the end of the intervention. 

Those with more education claim to learn 
more than those without across all technolo- 
gies, although, on average, the difference is 
small. Those with secondary education, on 
average, claim to increase their knowledge 
by 50 percent across all technologies, while 
those with no or primary education learn 
46 percent. For tithonia, the differences 
are somewhat greater. Those with second- 
ary education claim to have increased their 
knowledge to 64 percent (42 percent during 
the intervention and another 22 percent 
since). People with little or no education 
said they learned less, at 51 percent (only 
28 percent during the intervention, but a 
similar level of 23 percent since). But these 
differences are not especially high, indicat- 
ing that education level is not that important 
to knowledge acquisition with these tech- 
nologies, and that disseminating organiza- 
tions have been successful in reaching the 
less educated and thus more vulnerable res- 
idents. It also suggests that they have been 
effective in their training, with assistance 
from local disseminators. The average dif- 
ference of about 10 percent in knowledge 
gained between more and less educated peo- 

Table 9.2 Knowledge acquisition in 
agricultural topics from household survey 
in non-pilot villages (n= 361) 

Percentage of 

households with 

Agricultural topic 

knowledge gain 

Soil fertility assessment 


Use of fertilizer 


Use of manure 


Use of agroforestry 


Crop husbandry 


Livestock management 


pie is understandable in that the former have 
reaped the benefits of the cognitive changes 
associated with more education. 

Evidence of Knowledge Acquisition 
from Quantitative Surveys 

Regression analysis was done to examine 
the associations between dissemination 
method and major improvements in knowl- 
edge of several agricultural topics on the 
other. The dependent variables included 
improvements in agroforestry, fertilizer, soil 
fertility assessment, manure, livestock, and 
crop husbandry knowledge and were mea- 
sured as a dichotomous variable (and conse- 
quently a logit model was used). As indi- 
cated in Table 9.2, knowledge acquisition 
was common, ranging from 35 percent of 
households in the case of fertilizer to 70 per- 
cent of households for soil fertility assess- 
ment and use of manure. The fact that agro- 
forestry information does not rank in the top 
three in terms of prevalence attests to the 
fact that considerable dissemination activity 
occurs beyond SFR in the sites. 

We attempted two separate tests. The 
first was to evaluate the changes in knowl- 
edge according to site, but controlling for 
other household characteristics. This is only 
an indirect measure because the site dummy 
could be capturing other variables aside 
from dissemination differences, although it 
is not apparent that any of those would have 
changed over the period and have been 


related to increased knowledge. The results 
showed that fewer positive changes in agro- 
forestry knowledge were taking place in 
Mwitubi than in Bukhalalire, Shinyalu, 
Ugunja, and Rachuonyo. Fertilizer knowl- 
edge was improved less in Mwitubi and 
Rachuonyo than in Bukhalalire, Shinyalu, 
and Ugunjua. As for other types of knowl- 
edge (soil fertility assessment, crops, live- 
stock), there were hardly any further differ- 
ences between the sites, the exception being 
less perceived improvement in understand- 
ing soil fertility problems in Bukhalalire than 
in Mwitubi. 

The second analysis was to determine 
the impact of specific sources of informa- 
tion on SFR on changes in knowledge. We 
used direct contacts with different sources 
(yes or no) to proxy for whether the source 
was used by the household. A single house- 
hold may have received information from 
more than one source. Logit models with 
and without site variables were run. In terms 
of improved information in agroforestry, 
positive impacts were associated with direct 
contact with ICRAF, a nongovernmental 
organization (NGO), or a community-based 
organization (CBO). Contacts with exten- 
sion or other farmers did not have an effect 
on acquisition of agroforestry knowledge. 
Perhaps because the technologies are very 
new to the region, farmers were more wary 
of the knowledge and experience of exten- 
sion and other farmers. We also ran regres- 
sions for spillover effects in other types of 
agricultural knowledge, because all infor- 
mation providers attempt to address issues 
pertaining to other farming aspects. For all 
other types of knowledge, however, the 
only strong evidence we were able to find 
was that CBOs were able to impact posi- 
tively on knowledge gain in soil fertility as- 
sessment, use of manure, and use of live- 
stock. Other sources of information did not 
have an impact on knowledge acquisition 
in these other areas. This result is concern- 
ing in two ways. First, it suggests that ex- 
tension has not made a significant impres- 
sion. Thus, in spite of a liking for extension 

methods (see the qualitative analysis) exten- 
sion does not seem to reach many farmers 
with the methods. Second, the commonly 
formulated strategy of promoting farmer- 
to-farmer dissemination may prove to have 
a weakness in transmitting important con- 
textual and conceptual information about 
technology use. 

Differences in Gender and Wealth 
among the Focus Groups 

In the focus groups, there are not particu- 
larly large differences in knowledge gain 
by gender and wealth. Nonetheless, the few 
that exist are worth exploring. The clearest 
differences are between knowledge gain by 
wealth on commercial fertilizer and im- 
proved fallow. For fertilizer, non-poor partic- 
ipants report a 49 percent gain. Poor par- 
ticipants report a 39 percent gain. This is 
likely explained by the fact that commercial 
fertilizer is one of the only SFR technologies 
introduced that costs money. The converse 
is true for improved fallows. This receives 
the same scores but with "non-poor" and 
"poor" interchanged. Improved fallow serves 
more or less the same function as fertilizer, 
improving fertility, but does not cost any- 
thing, except forgone harvest. With regard 
to tithonia, the poor group was slightly more 
positive about their knowledge gain, at 57 
percent compared to non-poor at 51 percent. 
Women claimed they gained somewhat more 
knowledge on tithonia than men, at 57 per- 
cent versus 48 percent. In Sauri, although 
participants said that it is difficult for 
women to work with tithonia, women also 
said their knowledge gain was twice that of 
men for tithonia and improved fallows. 

In addition to these impressive gains in 
knowledge on the technologies, people 
report additional types of personal gains. 
Some mentioned that they gained yields as 
a result, but also confidence. What is most 
impressive is that groups of women and 
poor people mention the issue of confidence 
with greater frequency. Of the nine focus 
groups where this came up, six were 
women's and five were in poor groups. 

122 CHAPTER 9 


Project approaches and activities are con- 
sidered sustainable when local people per- 
petuate or support them, with minimal or 
no external assistance. This may be deduced 
on the basis of how local people support 
ongoing activities while external agencies 
participate in implementation or how they 
handle project activities and approaches, 
once the external organizations leave their 
communities. The current study uses both 
of those approaches for understanding and 
discussing sustainability. 

Project activity costs are an important 
challenge to the continuance of activities 
after external initiating organizations phase 
out of communities. The manner in which 
those costs are handled, therefore, constitute 
a significant way of weighing the sustain- 
ability of project activities and approaches. 
In some of the projects in western Kenya 
under study, farmers had demonstrated their 
willingness and ability to share in the proj- 
ect activity costs, provided that benefits are 
clear, and the disseminator is willing to do 
its part. Thus, according to our poor men's 
focus group in Mutsulio, "Farmers are ready 
to work with any organization on a cost 
sharing basis if only the organization is 
ready to stay in the village and tell farmers 
what will be benefited and steps to follow 
whenever problems arise." Farmers proved 
willing to provide locally available materials 
in support of respective projects. For in- 
stance, in Gongo and Sauri villages, farmers 
provided bricks, sand, and rocks while 
Africa NOW, the external organization ini- 
tiating a water project in the villages, pro- 
vided cement and paid constructors. Thus 
the external Africa NOW provided only 
items that the community could not raise 
in the circumstances of the time. In Gongo 
village, farmers provided a total of 33 acres 
to a KARI-sponsored project for demon- 
stration activities. In addition, in Mutsulio, 
Bukhalalire, and Mwitubi villages, farmers 
mobilized their colleagues for training ses- 
sions and barazas, and arranged venues as 

their in-kind contribution to projects. Over- 
all, farmers in all villages were aware that 
project activities cost time and funds, and 
had taken initiative in those costs in some 
way that they considered appropriate. 

In some instances, however, respondents 
in a few focus group discussions were not 
clear on how they shared in project costs, 
mainly because they looked at cost sharing 
as constituting monetary contributions. This 
is why a focus group discussion of poor 
men in Bukhalalire village reported that they 
gave monetary contributions to support dis- 
semination activities and were disappointed 
when they gave a bigger portion than 
ICRAF In the same vein, poor women of 
Mutsulio village said that farmers were "not 
ready" to cost share because they had no 
money. Some people saw cost sharing in a 
negative light. For example, poor women 
in Mwitubi stated that "farmers explained 
that a mere mention of the word money, 
that is, paying for something, is enough to 
send some members of social/farmer groups 
packing." But showing the strong will 
among farmers to sustain project activities, 
the poor women focus group of Bukhalalire 
village insisted: "as poor as we are, we 
may not make payment for most services 
but we are ready to do so whenever we 
can." Hence, it is the non-poor women 
focus group from Bukhalalire village who 
captured the in-kind aspects of cost sharing, 
arguing, "farmers also provide plots, labor 
and take the risk associated with experimen- 
tation on the farm." Cost sharing viewed 
from the in-kind perspective, therefore, 
leads to the conclusion that farmers met a 
significant part of the project costs in time 
and material, pointing to the communities' 
potentials in sustaining project activities 
and approaches in that sense. 

In terms of institutional sustainability, 
the study turned to the committees/groups 
that supported dissemination approaches 
and activities. The main issues for examina- 
tion were whether or not the said groups/ 
committees continued to exist after con- 


cerned external organizations had phased 
out of respective communities and if they 
existed, how they performed in terms of 
dissemination. The main finding was that 
most committees/groups set up or adopted 
for dissemination work by external organi- 
zations continued to exist after the latter had 
ceased to operate in the communities, al- 
though in some villages, groups had col- 
lapsed. In Bukhalalire village, the umbrella 
group was still active in dissemination in 
spite of the fact that KWAP, which initiated 
it, had left the community many years ear- 
lier. Poor men in the village said, "The um- 
brella group continues to disseminate to 
other villages beyond the catchment and 
giving briefings on its activities to schools." 
Thus the umbrella group disseminated 
much more widely than the MoARD had 
done through the catchment approach in 
earlier years. The umbrella group continued 
to use field days to disseminate to farmers, 
even after KWAP had left the village. In 
Mutsulio village, the committees were 
active and effective in dissemination and 
community members had not thought of 
making any changes in the ways they func- 
tioned, even after the external organization, 
KARI, had reduced contacts and ICRAF 
came and left the village. Focus group 
members said the committees in the village 
were motivated to work harder as the crop 
yields for the farmers they worked with kept 

However, a few committees, such as 
those for water set up by Africa NOW in 
Gongo village, had collapsed mainly because 
of mismanagement. Poor management, espe- 
cially of finances, had also kept some groups/ 
committees weak and ineffective in dissem- 
ination. In Mwitubi village, for example, the 
village committee had become inactive as a 
result of MoARD's withdrawal from the 
village, but also because of opaque manage- 
ment practices in the committee, especially 
with regard to finances. In Gongo, Sauri, 
and Muhanda villages, most groups were 
active, except for some groups that had 

collapsed in Gongo and Muhanda, due to 
financial mismanagement or thievery. It is 
worth noting, however, that mismanagement 
is not a pervasive problem; these are the 
exceptions. For example, in Bukhalalire, 
non-poor women said that "mismanage- 
ment is not a constraint to participation in 
these groups because there is no money to 
mismanage, but it is a minor problem." 

In all focus groups, the local adminis- 
tration came across as important to sustain- 
ability. The chiefs, their assistants, and vil- 
lage elders mobilized farmers for barazas 
and other dissemination sessions, and some- 
times disseminated directly to them. This is 
important for sustainability because after 
the external organizations left the villages, 
the chiefs remain in their influential posi- 
tions, and have the potential to continue to 
convene barazas that support dissemina- 
tion. Although this is potentially an impor- 
tant source of sustainability, however, it is 
not sufficient, as chiefs sometimes complain 
that people do not attend barazas anymore. 
The major challenges to sustainability may 
be adduced at two levels: at the level of cost 
sharing and at the level of institutions. At 
the level of cost sharing, the major problem 
relates to the high levels of poverty in the 
project areas, which makes it difficult for 
farmers to continue to fund activities and 
groups sufficiently after external organiza- 
tions phase out. In some cases, farmers said 
they were willing to share costs, but they 
would first need to see the financial benefits 
of adopting the technologies. At the institu- 
tional level, the local administration was in- 
volved in dissemination ad hoc and without 
undergoing any structured training on dis- 
semination. Therefore, although the admin- 
istration was said to be involved in dissem- 
ination, the content may not have been in 
line with the desires of farmers; however, 
given the powerful position of that adminis- 
tration in the lives of the people, they could 
not question it. Also at the institutional 
level, poor leadership and mismanagement 
of funds hampered the continued vibrancy 

124 CHAPTER 9 

of groups/committees after external organi- 
zations left the communities. The major rea- 
son that respondents cited for the collapse 
or inactivity and ineffectiveness of groups 
was mismanagement of funds or poor lead- 
ership. For institutional sustainability to be 

achieved, future projects should focus some 
more on leadership and management train- 
ing, to provide safe grounding for project 
activities and approaches after external or- 
ganizations phase out. 


Conclusions and Recommendations 

This chapter serves three main objectives. The first is to discuss the advantages of the 
research methodological approach and remaining challenges. The second is to high- 
light the key empirical results of the impact assessment study. The last is to explore 
more broadly alternative options for poverty alleviation in western Kenya and the role of soil 
fertility investment. 

Research Methods 

The study used a number of methods/approaches that shaped the research process and the 
findings. The important aspects are the following: 

• Use of a sustainable livelihoods (SL) framework 

• Integration of qualitative and quantitative methods 

• Dialoguing across different institutions and stakeholders 

• Arm's length data collection 

• Viewing poverty from multiple perspectives 

Use of a Sustainable Livelihoods Framework 

The SL framework was used in the formulation of the final proposal and in identifying the key 
research questions. It provided a cross-disciplinary language that allowed for the articulation 
of an integrated research design. This was attractive. As it turned out, the research team was 
conversant in a number of paradigms and methods that included similar concepts, and it is 
not clear that the SL framework was necessary to achieve similar results. Still, the framework 
ensured that many key issues and relationships were not overlooked. The team introduced 
other concepts not included or explicit in the framework as needed. In terms of research ques- 
tions, it was noted that the SL framework, although quite comprehensive, did not provide 
guidance as to the major direct relationships among variables. As a consequence, all variables 
are related to one another and this then tends to result in the formulation of similar, overlap- 
ping, or duplicated research questions. 

In the implementation of the research, the SL framework was rarely explicitly discussed, ex- 
cept in the formulation of the research questions and during certain forums where these ques- 
tions and preliminary findings were reviewed. This is less a criticism or compliment to the 
framework than a reflection of the fact that the research team had agreed on the importance of 
a comprehensive and diversified research approach. 

Integration of Qualitative and Quantitative Methods 

As the research questions were being developed, the research team noted the particular 
contribution that could be made from qualitative and quantitative methods. For most of the 


126 CHAPTER 10 

questions, it was clear that qualitative and 
quantitative methods could complement one 
another. As much as possible, checklists and 
questionnaires were formulated to provide 
insights into common issues/questions. Cer- 
tainly, the integration was useful for under- 
standing different types of information — 
quantitative results led to identification of 
general patterns and qualitative results to 
help understand processes of adoption 
choices, information flows, and impacts. 
Quantitative results give a much stronger 
sense of representativeness, given the large 
numbers, but qualitative data were essential 
for uncovering issues related to culture, nor- 
mative frameworks, and social dynamics. 

The integration succeeded in issues such 
as meanings of poverty, adoption, and dis- 
semination issues. It worked less well with 
impact analysis, mainly because the quan- 
titative research had a baseline as a guide, 
but was limited to a small set of indicators, 
while the qualitative research did not bene- 
fit from a baseline, but was broader in its 
scope. Although a great deal of complemen- 
tary and supporting information is available, 
true integration requires the researchers to sit 
together and compare/contrast results. Lim- 
itations on time and funding, and the timing 
of the last survey, resulted in the team not 
being able to spend sufficient time to jointly 
analyze the research results. The report is 
therefore too compartmentalized into results 
from different methods. 

Another issue was the use of numbers 
generated from PRA exercises. These were 
quite useful in understanding relative as- 
sessments of different institutions, methods, 
and knowledge acquisition within villages. 
They were less useful than they could have 
been, however, and difficult to compare and 
analyze across villages, because of the dif- 
ferent meanings attached to numbers, and 
the fact that different categories were iden- 
tified across villages. This could have been 
controlled by standardizing categories, but 
would have undermined the participatory 
nature of the exercise, stifling generation 
of local categories. Still, there was enough 

comparability across categories to allow 
for some meaningful comparison, albeit in 
broad strokes in some places. 

Dialoguing across Different 
Institutions and Stakeholders 

A key aspect of the research process was 
the stakeholder meetings that helped to 
plan and review the research. The launch- 
ing stakeholder workshop in western Kenya 
was particularly valuable to some members 
of the research team who were unfamiliar 
with the research setting. Having issues 
raised by residents of the region added a de- 
gree of objectivity to the research design that 
may not have been as defensible had the de- 
sign been driven by ICRAF and its partners 
alone. It also increased the practical rele- 
vance of the results, and ensured an audi- 
ence for the results. In addition to the initial 
workshop, there were several meetings 
among the research team members, which 
also involved other stakeholders, such as 
development agents. These meetings helped 
to plan specific components of the research, 
such as the focus groups on dissemination 
strategies that included representatives of 
organizations whose methods we were eval- 
uating. Some of these same stakeholders 
acted as sources of data for the project as 
well, providing some data triangulation. 

Arm's-Length Data Collection 

The qualitative and quantitative fieldwork 
was undertaken almost exclusively by per- 
sons not attached to the project. The quali- 
tative fieldwork was supervised by Ph.D.- 
and M. A. -level faculty sociologists, and 
carried out by M.A.- and B. A. -level soci- 
ologists, all of whom had current or former 
connections to the University of Nairobi 
and Wageningen University. They were 
supported by the "Social Analysis Team," 
composed of sociologists from IFPRI and 
Wageningen University. The exception to 
this was assistance with organizing (but not 
facilitating) some focus group discussions 
from an ICRAF social scientist. There were 
two reasons for using independent (non- 


ICRAF) researchers. First, we wanted the 
fieldworkers to be viewed by villagers as 
completely detached from ICRAF. ICRAF 
has had mixed experiences in the past with 
obtaining open responses from farmers in 
the pilot villages. They have often hidden 
problems with the technologies from ICRAF 
scientists unless specifically probed. Thus, 
we felt that it was essential to de-link the 
collection and analysis of qualitative data 
from ICRAF. It is not clear that this worked 
that well in practice, as ICRAF is well 
known in the area and much of the inquiries 
were related to agroforestry. Nevertheless, 
the long period of fieldwork built up rela- 
tionships between fieldworkers and respon- 
dents that produced candid replies. 

The second reason for employing in- 
dependent researchers was to maximize the 
objectivity of the impact assessment pro- 
cess, for example, reducing conflicts of in- 
terest with respect to reporting negative 
results where they may occur. With respect 
to qualitative data analysis, ICRAF was not 
involved in the initial analysis but did re- 
view it and provide feedback to the analysts. 
While the use of independently contracted 
social analysis researchers did help with per- 
ceived and real objectivity, this also meant 
that internal capacity for undertaking social 
analysis was not built within ICRAF. Never- 
theless, ICRAF has built relationships for 
new potential collaborators in the future. 

The issue of the independence of eval- 
uators was less of a concern with respect 
to survey enumerators, but we nevertheless 
felt it advisable to use students and other 
collaborators to collect these data as well. 
Because of the complicated nature of the 
survey questions, the same enumerators who 
collected the baseline information on con- 
sumption and expenditures were used again 
to ensure quality and consistency, and to 
make households more comfortable in dis- 
cussing the topics. In the non-pilot villages, 
ICRAF was involved only in setting up the 
logistics. Students from Belgium supervised 
the collection of data in three villages while 
a student from the University of Nairobi 

supervised data collection in the other three 
villages. Both students and ICRAF scien- 
tists participated in the data analysis, but 
because of strict deadlines and complexity 
(with respect to the SFR systems and the 
econometrics), the ICRAF project leader 
contributed the vast majority of the quanti- 
tative analysis in this report. However, this 
was guided by mutually generated questions 
and oversight from IFPRI reviewers and the 
independent advisory committee. 

Viewing Poverty from 
Multiple Perspectives 

As noted earlier in this report (Chapter 4), 
poverty is a slippery concept. Yet the task 
was to see how, if at all, the agroforestry 
technologies impacted on the poor. So the 
team was forced to come to grips with how 
to assess who is poorer compared to others. 
Rather than devising a single qualitative 
or quantitative classification, the team was 
open to alternative views and ways of 
comparing poverty levels across households. 
This approach seems to best fit with reality 
in that poverty is dynamic — many house- 
holds are vulnerable to poverty and are en- 
gaged in a range of survival strategies. In an 
effort to maintain consumption levels, many 
households will resort to working off-farm 
(and potentially jeopardizing yields) or sell- 
ing assets. By looking at only one of these 
dimensions of poverty, important changes 
or effects could be missed. It is important to 
look at the many manifestations of poverty 
for seeking solutions as well. Certain types 
of poverty indicators may move together, 
while others may not. For example, we found 
that expenditure and consumption changes 
over time were quite similar, but asset port- 
folio changes behaved somewhat differently. 

Highlights from 
Empirical Findings 

1. While there is no doubt that poverty is 
pervasive in western Kenya, distinguishing 
the poor from the non-poor is not straight- 
forward. Poverty is a slippery concept and 

128 CHAPTER 10 

people often do not accept being labeled as 
poor. When pressed, people will admit that 
poverty implies the lack of certain basic 
needs. The study used a variety of methods 
to assess poverty levels, including quantita- 
tive measures from surveys, enumerator 
ratings, and farmer self-assessments. These 
produce different outcomes so that which 
households are classified as very poor de- 
pends on the criteria used. 

2. Welfare or livelihood outcomes are 
worsening for many households. There was 
a general deterioration in welfare indicators 
during the period of study. This holds true 
for assets, expenditures, and food consump- 
tion. Particularly striking was that house- 
holds with relatively high welfare indicators 
in the initial period suffered the greatest 
losses. This is attributable partly to the large 
number of adverse shocks affecting house- 
holds and the cultural obligations felt by all 
community members (e.g., the wealthier 
households contribute animals for slaughter 
at funerals). 

3. SFR technology interventions imply 
assumptions about the role of agriculture in 
people's livelihoods that have varying va- 
lidity. The role of agriculture in people's live- 
lihoods is determined by economic circum- 
stances, cultural, normative frameworks, and 
social identities. The assumption that poverty 
can be reduced through farming is not 
necessarily reflected in the investments in 
livelihood activities made by people in the 
region. Their decisions are embedded in 
their economic circumstances (including 
assets and institutional environment), cul- 
tural, normative frameworks, and social 
identities. Decisions about agricultural in- 
vestments are also shaped by "structural" 
phenomena such as the output to input price 
squeeze on agriculture that does not guar- 
antee adequate return to human and physi- 
cal capital investments. In western Kenya, 
farmers are very aware of this squeeze in 
making livelihood decisions. While re- 
searchers may evaluate agroforestry in 
terms of its role in generating agricultural 
production, rural people will be assessing its 

ability to contribute to the variety of objec- 
tives they pursue. 

4. Households do realize the impor- 
tance of SFR — and there have been many 
human capital impacts. Both the qualitative 
and quantitative research found significant 
knowledge acquisition taking place, not 
only for agroforestry methods, but also for 
general soil management and farming prac- 
tices. People valued this information and 
have often put it into practice. 

5. Farmers like to be exposed to mul- 
tiple dissemination opportunities and meth- 
ods. The dissemination analysis found that 
farmers appreciated some efforts of almost 
all disseminating organizations and certain 
aspects of the many different methods 
tried. They particularly appreciated direct 
contact and field observation methods. In- 
formation flows were not guaranteed, how- 
ever, because individuals may not be able to 
make scheduled meeting times and different 
methods benefit some social groups more 
than others. Thus, the poor favor being able 
to access information through a variety of 

6. Social status and social relationships 
within villages affect outcomes of different 
dissemination methods. Although charac- 
teristics of SFR affect whether people adopt 
or not, aspects of the dissemination process 
also affect adoption. The dissemination 
analysis found that the main feature of most 
dissemination approaches — group-based 
methods — can strengthen human and social 
capital, and farmers of different social status 
have benefited from them. This analysis also 
found, however, that group-based approaches 
can also disadvantage farmers of lower so- 
cial status and women who are less likely 
to participate in or have key positions 
in groups. However, women's groups have 
worked well for women. Furthermore, the 
dissemination analysis and case studies 
found that the use of adaptive research 
farmers is necessary, but it also generated 
social tensions because of the amount of 
attention selected individuals received from 
outsiders. These findings reinforce the con- 


elusion that use of a variety of methods are 
best, and point to the importance of under- 
standing local social dynamics in designing 
dissemination interventions. 

7. Sustainability of dissemination struc- 
tures and processes is possible but tenuous. 
Sustainability of dissemination structures 
and processes has proved to be possible, 
but challenging, because of problems en- 
countered by groups, limited capacity of 
local administration, social dynamics within 
villages, and limited cost-sharing ability. 
Monitoring would help to pick up these 
problems so that resolutions can be sought 
where possible. 

8. The poor are adopting SFR strategies 
at rates similar to those of the non-poor. In 
villages where information on SFR was dis- 
seminated several years ago, adoption rates 
are not outstanding but they are encourag- 
ing. About 20 percent of all farmers are 
using the technologies on a regular basis (a 
similar percentage among the poor) and a 
sizable percentage of farmers are newly test- 
ing them. Using different datasets and non- 
land wealth measures, it is found that rates 
of use are similar across most wealth/ 
poverty indicators. Other types of soil fertil- 
ity measures such as manure and fertilizer 
are more clearly linked to non-land wealth 
indicators. This does not mean that SFR 
strategies are equally likely to be adopted 
across all levels of asset or resource holdings. 
Biomass transfer was particularly related to 
the pool of household labor and improved 
fallows to sufficient land holding size. 

9. Adoption at an early stage is at low 
levels of intensity. While an encouraging 
number of households are using or testing 
the SFR practices, the sizes of plots on 
which they are applied remain small. It is 
not yet known whether this is indeed an op- 
timum ceiling, or a consequence of the early 
stage of dissemination, in which villages 
are learning to adjust to a significant with- 
drawal of support from project personnel. 

10. New technologies transform or 
create/introduce new existing social rela- 
tionships in (rural) societies. New tech- 

nologies and the methods of dissemination 
have created social tensions and contribute 
to (more) conflicts. The nature of such ten- 
sions complicates an assessment of technol- 
ogy interventions and has much more to do 
with the way technologies are/were intro- 
duced and thus has been of more concern in 
the pilot villages established by ICRAF 

11. SFR does significantly raise crop 
yields in most cases. The best test of this 
was from longitudinal farmer managed tri- 
als that showed significant yield and returns 
to labor gains. Respondents in the case 
studies and formal surveys also consistently 
report very significant increases in yields 
(>100 percent) from the use of SFR prac- 
tices. Not all farmers benefit from SFR, 
however; there is variation in performance 
and these are difficult to isolate and quantify 
from farmer recall surveys. 

12. SFR on its own cannot bring about a 
turn in poverty reduction. This conclusion is 
drawn from the body of impact assessment 
work. Despite the fact that SFR is being 
used by a number of poor households and 
has an impact on yields, its impact at the 
household level is modest because of the 
small land sizes under SFR and because 
the weak rural economy is not conducive 
for investment and development. This means 
that technological innovations alone are 
likely to have a limited short-term impact. 
Poverty alleviation interventions must en- 
compass other sectors as well. 

Implications for Poverty 
Reduction in Western 
Kenya: The Way Forward for 
Soil Fertility Replenishment 

Pathways out of poverty are varied and 
highly uncertain. It seems that in order for 
widespread poverty alleviation to take place, 
many components of the rural socioeconomy 
need to be functioning well. This means that 
many agricultural enterprises and their 
markets need to be promising, the nonfarm 
economy needs to be growing, human and 
animal health diseases need to be kept at 

130 CHAPTER 10 

bay, information flows need to function well, 
and general pessimism needs to be replaced 
by optimism. Macro economically, the per- 
formance of the non-agricultural sector has 
been poor, offering few opportunities for 
those who aim to diversify from agriculture 
into other sectors of the economy as labor 
migrants or entrepreneurs. In the short term, 
improvement in the lives of rural house- 
holds will have to evolve around unlocking 
land-based agricultural activities and op- 
portunities. Smallholders in central Kenya, 
with nearly the same sized farms, generate 
more than three times as much agricultural 
revenue as do farmers in western Kenya. 
This comes about partly because of extraor- 
dinary yields, but more as a result of agri- 
cultural enterprise choice, in which high- 
value enterprises such as tea, dairy, coffee, 
macadamia, and other horticulture products 
are common. Given that some of these en- 
terprises enjoy growing export markets (e.g., 
tea and macadamia), agriculture also seems 
to offer some promise throughout medium 
to high potential areas of Kenya. This is par- 
ticularly so in western Kenya, because soils 
are deep and well drained and rainfall pat- 
terns are arguably the best in all of Kenya. 
But realizing this vision for agriculture 
has many obstacles. Prices for agricultural 
commodities are low and agriculture contin- 
ues to be subjected to climate risks. Most of 
the profitable agricultural enterprises require 
capital investment and poor households 
simply do not have sufficient financial re- 
sources. They may try to start more mod- 
estly, but even that process is problematic. 
Household responses to the HIV/ AIDS pan- 
demic are creating a drain on financial and 
human capital resources, the latter through 
illness, care taking, and attendance at lengthy 
funeral rituals. Small land sizes in turn limit 
the amount of diversification that households 
are willing to undertake. A major challenge 
is that irrespective of the direction taken, 
agriculture needs to proceed in such a way 
that monetary costs are kept at low levels. 
The scenario of labor intensification is the 
most likely outcome, as it can generate 

opportunities for the existing pools of rural 
labor and reduce the monetary costs of 

Within agriculture, poor households 
can take initial steps by building on crops/ 
enterprises that they already have. The strat- 
egy under consideration in this study was 
a relatively safe one of increasing yields of 
the basic staples of maize and bean. These 
are safe because increases in production can 
be consumed on-farm and markets always 
exist for these commodities. However, their 
value is low, so that safety is traded for only 
modest income boosts. 

More substantial income increases may 
come from increasing investment in poultry, 
woodlots, or vegetables, which are also 
found among the poor. In order for these 
enterprises to generate sustainable incomes, 
the greater socioeconomy must be function- 
ing well. Poultry is highly vulnerable to 
disease, vegetables require outlays of scarce 
capital, and woodlots require a waiting pe- 
riod in which income must be derived from 
other sources. These risks and investment 
costs are more easily borne through sur- 
pluses generated from other enterprises or 
livelihoods. Promoting diversity of options 
will continue to be important. Pathways out 
of poverty based on specialization (in tea, 
and so forth) is interfacing with the cultural 
frameworks of local, rural people, who have 
been trying to improve their lives through 
diversification. Further, households are likely 
to continue to allocate significant amounts of 
land and labor resources to food-producing 
crops. Of course this may change in the near 
future. But some will adapt to the new cir- 
cumstances; others will not. 

Even if progress is made, the study has 
clearly shown that households can easily 
slip back into poverty conditions. In addi- 
tion to generating production and income, 
there is need for insurance through invest- 
ment in risk-buffering assets. Without these 
assets, the numerous adverse shocks can 
easily impact on household expenditures 
and consumption levels. Markets that enable 
households to build and divest assets when 


necessary do exist. What is more problem- 
atic are the sociocultural pressures that force 
the more well-off households to sacrifice 
assets for funerals and other hardships. The 
well-off also tend to be targets of theft, in 
both on-farm and off-farm activities. Such 
social phenomena can discourage the 
buildup of the very assets that are needed to 
reduce vulnerability to poverty. 

What is the future for agroforestry in all 
of this? The soil fertility systems being dis- 
seminated are a useful option for farmers. 
There are cases in which this one type of 
technology seemed to trigger more sustained 
development. There are clear limitations to 
the use of improved fallows and biomass 
transfer, however. Improved fallows are less 
likely where farms are very small and where 
other perennial crops are found. There are 
similarly some limitations of space for grow- 
ing organic biomass, but it seems to be a 
useful option, especially as part of a strategy 
to intensify production of high-value crops. 

Agroforestry does combine aspects of 
indigenous practice and knowledge with 
advanced science. The SFR systems enable 
farmers to be able to improve soil fertility 
without reliance on markets and with little 
expenditure of the most precious production 
factor, money. This can become real only 
by theoretically rethinking the notion of 
resources in two ways. One is to move be- 
yond the view that resources are a given, to 
be used as designed by the developing sci- 
entists. It is more appropriate to imagine 
resources as unfolding. Resources unfold 
in and through practice and this involves 
learning by doing. That resources unfold is 
reflected in the processes of adaptation one 
witnesses in the field. Farmers often change 
the original design when it does not fit with 
their ideas. Sometimes farmers rework a 
technical design in a way that the original 
version cannot be detected. The notion of 
unfolding is also important in that sometimes 

resources seen as relevant by scientists are 
initially seen as irrelevant by farmers, who 
only later may discover their relevance. 

The case material and our analysis have 
indicated a second direction in which re- 
sources can be rethought: resources need to 
be understood not only as material objects 
and artifacts, but also as a social relation- 
ship; social relationships are embedded in 
technology. This means paying attention to 
the relationships that emerge between tech- 
nology designers and agents of extension, 
on the one hand, and the "beneficiaries," on 
the other. It also means paying careful at- 
tention to the types of individuals and house- 
holds who are likely to be able to under- 
stand and adopt agroforestry for improving 
soils. Theoretically and empirically so far, 
the technologies do not seem to strongly 
discriminate between the rich and the poor 
as do other technologies, because of their 
low cash requirements, although some 
minimum amount of land is necessary for 
improved fallows and similarly some ability 
to command labor for biomass transfer. 
Also, both men and women are using them. 
Where it has led to social tensions seems to 
relate more to the manner in which infor- 
mation about the technology is disseminated. 
Many potential users of the technology may 
not receive information or become alienated 
from trying if dissemination methods are 
not considered to be equitable. These tech- 
nologies carry certain requirements, not the 
least of which are those related to knowl- 
edge. Systems for diffusing information 
over wide areas are critical and the fact that 
agroforestry for SFR has not yet been well 
established in extension systems means that 
considerable work is required in building 
capacity in national programs. For ICRAF, 
it should see its role in identifying the key 
research gaps that constrain the possible bas- 
ket of options that extension and other devel- 
opment organizations can bring to farmers. 


Results and Tests from 
First-Stage Regressions 

Results from First-Stage Regressions to Predict 
Improved Fallow Area and Number of Seasons 
Practicing Biomass Transfer 

Table A.1 OLS regression results for improved fallow area 



Df MS 



14 729038.54 



84 346538.401 



98 401181.278 







[95% Conf. interval] 












































































































F(14, 84) 


Prob > F 


R 1 


Adj. R 2 


Root MSE 




Table A. 2 Tobit regression results for improved fallow area 






P> \t\ 

[95% Conf. interval] 













































































































(Ancillary parameter) 



LR x 2 (14) 


Prob > x 2 


Pseudo-R 2 


Log likelihood 


Table A. 3 OLS regression results for number of seasons practicing biomass transfer 







14 2.13910788 



84 2.05425643 



98 2.06637807 







[95% Conf. interval] 












































































































F(14, 84) 


Prob > F 


R 2 


Adj. R 2 


Root MSE 



Table A. 4 Tobit regression results for number of seasons practicing biomass transfer 






P> \t\ 

[95% Conf. interval] 













































































































(Ancillary parameter) 



LR X 2 (14) 


Prob > x 2 


Pseudo-R 2 


Log likelihood 


Tests on the Selection of Instruments for 
First-Stage Regressions 

1. Correlation tests between predicted and actual values 

1A. Improved fallows 

Correlation from OLS first-stage regression: 
Correlation from Tobit first-stage regression: 

IB. Biomass transfer 

Correlation from OLS first-stage regression: 
Correlation from Tobit first-stage regression: 
All correlations are significant at the .01 level. 




r= .36 


F-tests and % 2 tests for first-stage regressions 

2A. Improved fallows 

Significance of F-stat from OLS first-stage regression: 
Significance of % 2 stat from Tobit first-stage regression: 

2B. Biomass transfer 

Significance of F-stat from OLS first-stage regression: 
Significance of % 2 stat from Tobit first-stage regression: 





3. Tests for overidentification 

Significance of n*R 2 from regression of impact residuals (actual values of impact indica- 
tors less their predicted values) on all RHS variables, including the instruments (with de- 
grees of freedom of number of instruments (7) less the number of endogenous RHS vari- 
ables (2) = 5). 


3A. Change in expenditure per capita regression: 

% 2 = .9306, overidentification hypothesis is rejected. 
3B. Change in protein regression: 

% 2 = 2.6532, overidentification hypothesis cannot be rejected. 
3C. Change in iron regression: 

X 2 = 2.8512, overidentification hypothesis cannot be rejected. 
3D. Change in energy regression: 

% 2 = 5.6034, overidentification hypothesis cannot be rejected. 
3E. Change in assets regression: 

X 2 = 2.2466, overidentification hypothesis cannot be rejected. 

4. Durbin-Hausman-Wu tests on significance of agroforestry residuals (actual values of 
agroforestry variables less their predicted values) in regression with impact indicators as 
dependent variables. 

Tests for significance of residuals of first-stage regression on dependent variables. 

Regression Improved fallows residuals Biomass transfer residuals 

Change in expenditures per capita + (p = .04) + (p = .09) 
Change in protein n.s. n.s. 

Change in iron + (p = .06) - (p = .10) 
Change in energy n.s. n.s. 

Change in assets n.s. n.s. 

n.s. = not significant. 

5. Hausman specification tests comparing estimates on agroforestry variables from OLS 
and 2SLS procedures. 

Significance levels for differences between estimators in the two procedures. 

Regression Improved fallow predicted value Biomass transfer predicted value 

Change in expenditures .082 .185 

per capita 

Change in protein .272 .439 

Change in iron .130 .159 

Change in energy .918 .256 

Change in assets .914 .403 


Six Village-Level Case Studies of 
Dissemination Processes 

Michelle Adato, Mary Nyasimi, 
Christopher O'Leary, and Eduard Ontita 


elow are detailed village-level case studies for the six villages where dissemination 
processes were studied, using focus group and PRA methods. These findings are 
summarized in Chapters 8 and 9. 

The Village Approach in Sauri 

As the only Luo village where International Center for Research in Agroforestry (ICRAF) was 
the primary disseminator among the six villages studied, Sauri is of special interest for this 
study. This dissemination approach attempted to be simultaneously broad and sensitive to 
local culture. ICRAF believed that this was a means by which they could avoid the pitfalls of 
providing irrelevant or unusable information. It was largely successful. People in Sauri gen- 
erally think more highly of the organization than they do of others that do similar work in the 
village. The organization, however, has not lived up to all of its goals for accessibility. 
Nonetheless, the infrastructure they set up remains strong. Sauri's participants are unanimous 
in their positive assessment of the SFR technology, and the dissemination methods. 

ICRAF, along with its partners, designed their work to be adopted in this area of great 
need. Sauri experiences significant ecological problems. Although the area has sufficient rain- 
fall, this alone is inadequate. Local soils are notably deficient in key nutrients, especially nitro- 
gen and phosphorous. In addition, local population density rests at one of the highest levels in 
the world for agricultural populations at about 1,000 people per square kilometer. To top 
it off, the population growth is 3.4 percent per annum, putting even more of a crunch on local 
farmers. These factors make Sauri an important test case for any method that would increase 
soil fertility. It is an even more important process here as government funds for more capital- 
intensive development are not forthcoming. 

Assessment of Disseminating Organizations 

By and large, local assessments of ICRAF show that people hold the organization in high re- 
gard. All of the four focus groups rank it at 70 percent in terms of usefulness and importance. 
While this number itself is only a little higher than the other community where ICRAF is most 
active (Mwitubi), what is perhaps more important is that there is almost no disagreement re- 
garding ICRAF. Variations across the four focus groups are less than 10 percent, making this 



case one of the most unanimous of any as- 
sessment. Furthermore, no criticisms were 
made about ICRAF in the focus groups. 
One description of ICRAF was "motivat- 
ing." In other communities, many complain 
about insufficient time outside organiza- 
tions spend in communities, even when they 
spend far longer in a place. In Sauri, people 
noted that one year was short but sufficient. 
Poor men were dissatisfied with groups that 
left after five years of activity in the past. 
But, when asked about ICRAF, they quickly 
retracted this statement, saying that one year 
was plenty of time. 

The assessment of organizations active 
in the area was varied, indicating that peo- 
ple assess other things than technology when 
evaluating organizations. Kenya Forestry 
Research Institute (KEFRI) was ICRAF's 
direct research partner here. In addition, 
the Ministry of Agriculture and Rural De- 
velopment (MoARD) entered the commu- 
nity a little later, and sought to disseminate 
similar technologies. Finally, CARE was 
active in the community, and used schools 
as a means to spread knowledge. These other 
organizations scored less than half as well as 
ICRAF. People disagreed more on MoARD's 
effectiveness, with poor men and non-poor 
women scoring MoARD about twice as high 
as the other groups. KARI did not seem 
to make an impression in this village. Par- 
ticipants are generally positive about the 
technology disseminated by CARE. In fact, 
all groups were grateful for the trees they 
planted. These are used to pay school 
guards, buy chalk and blackboards, and im- 
prove buildings. Yet, in the scoring, CARE 
was credited with about one-seventh the im- 
portance of ICRAF. Resident's feelings to- 
ward ICRAF stem from their dissemination 
methods and the resultant knowledge gains. 

Teaching Methods 

ICRAF's stated goal was to disseminate via 
local groups that would then conduct field 
days, trials, training at meetings, and to show 
an example that other farmers could emulate. 
Local people would spread these technolo- 

gies via informal means such as conversa- 
tion as well. But farmers ranked conversation 
as a means of learning far lower than in 
other villages. This is most important in look- 
ing at poor men's assessments of informal 
conversation as a way to gain information. 
In the communities as a whole, poor men 
ranked conversations at 50 percent in terms 
of their usefulness and importance. In Sauri, 
the number is 7 percent. Similarly, focus 
groups in other villages ranked conversa- 
tions at 15 percent. Here, the average came 
to about 10 percent. Similarly, small group 
meetings were ranked at 10 percent (in non- 
poor men) and 35 percent (in poor women) 
in other communities. Here, only poor men 
and women mention that they learn anything 
about SFR technology from social groups. 
Poor women put their importance at 15 per- 
cent, less than half of the average. People in 
Sauri do not believe that they are getting 
much information from groups, and groups 
are less important for dissemination than 
ICRAF intended. Furthermore, people in all 
groups complained that the groups distrib- 
uted information about technology unfairly. 
Given that some of ICRAF's intended 
methods did not have the effect intended in 
Sauri, the reason for ICRAF's high status 
in participants' minds emerges as an im- 
portant question. The answer lies in formal 
extension methods. Most notable in Sauri 
assessments of dissemination methods is 
that barazas and tours emerge about twice 
as high as in the study communities gener- 
ally. Barazas here were ranked at about 
40 percent, and tours at about 20 percent in 
all Sauri focus groups. Perhaps most surpris- 
ing is that poor men scored tours at 25 per- 
cent in Sauri but 10 percent, on average, 
across all villages. These numbers combined 
suggest that formal methods are more ac- 
cessible to the poor in Sauri than they are in 
the communities as a whole. People also hold 
even more formal methods in high regard. 
A specific part of ICRAF's dissemination 
method was posters that describe technolo- 
gies, extension agents, and even field tech- 
nicians in some villages that visit farmers. 


Although these did not show up in the PRA 
exercise, all focus groups positively com- 
mented on both of these. 

Indeed, it is difficult to argue with 
ICRAF's effectiveness at disseminating tech- 
nology. Average farmer assessments of their 
knowledge gain in Sauri were about equal 
for all disseminated technologies when 
compared to other communities. The biggest 
difference appears in rock phosphate, which 
was higher in Sauri (44 percent versus 30 
percent, on average). Poor men ranked low- 
est of the four groups with regard to knowl- 
edge gain on all technologies except rock 
phosphate and improved fallows. Especially 
notable in this regard is tithonia, where 
poor men reported a 25 percent knowledge 
gain, while both women's groups said they 
gained over 65 percent. But, this does not 
stop with poor men. In contrast to many other 
locations, both men's groups reported less 
knowledge gain on tithonia than women of 
similar social status. Nonetheless, this rela- 
tively low assessment does not take away 
from the fact that people in Sauri believe 
they are gaining more than their counterparts 

Local Dissemination 

The most interesting and diverse aspect of 
ICRAF's methods involved use of local in- 
stitutions in dissemination. Here, ICRAF 
tried to use local groups for a variety of 
reasons. One way to increase adoption of po- 
tentially helpful technologies is to use local 
groups. This is the keystone of ICRAF's ap- 
proach, which is covered in the main body 
of this report. ICRAF conducted a study 
of village networks and groups in selected 
villages and concluded that virtually all 
Luo, including those in Sauri, belong to one 
social group aside from their immediate 
family. Examples of others common in the 
area include church organizations, women's 
groups, and funeral associations. ICRAF 
assumed that these groups are especially 
effective ways to reach other farmers and 
effect change since they are important fo- 
rums where people talk about politics and 

innovation, among other things. Thus, work 
with such a group can be a way to facilitate 
dissemination. Specifically, these groups 
were used to then filter the information 
down through smaller groups until it would 
reach farmers. This had a number of advan- 
tages, such as spreading already stretched re- 
sources over a greater area, and eliminating 
linguistic problems associated with multi- 
lingual societies such as this one. 

This approach also attempted to 
strengthen these groups. Through this, they 
hoped to lay the groundwork for later tech- 
nology dissemination. It would also allow 
greater farmer input, as stronger groups 
would presumably be better at arguing for 
their own ideas. Groups were used by dif- 
ferent organizations for different purposes. 
In Sauri, they were taught technologies such 
as improved fallows, composting, and plant- 
ing trees for fruit, wood, and medicine. They 
were also trained in bookkeeping, leader- 
ship, and proposal writing. Furthermore, they 
hoped to expand to other areas of training 
such as beekeeping, microcredit schemes, 
and more complex fruit growing. 

Local assessments of these locally based 
approaches supports their effectiveness. 
Participants ranked barazas as the most 
important and useful form of dissemination. 
In fact, barazas scored higher among all 
groups, with the exception of non-poor men. 
Poor men and non-poor and poor women 
ranked barazas at 28 percent, 40 percent, 
and 40 percent, respectively. All of these are 
higher than their scores in other communi- 
ties. In contrast, non-poor men ranked other 
farmers at 45 percent and schools at about 
33 percent. Considering that the analysis 
above shows that barazas in Sauri may be 
particularly appealing to the poor, and that 
non-poor men probably have more active 
social networks where they can learn about 
new technologies, we could expect them 
to assess other farmers as more important 
than barazas. Similarly, they were the only 
group that mentioned parent meetings as a 
way they learn information. This probably 
indicates that non-poor men are most active 


in these groups, and therefore the most likely 
to know about, discuss, and use school pro- 
grams. Finally, no group mentioned funerals 
as a means to disseminate information, and 
only a few mentioned churches. Finally, only 
poor men mentioned children as a local dis- 
seminator. The last is not especially surpris- 
ing, as schools were not such important 
sources of information in Sauri. The relative 
unimportance of funerals and churches as 
sources of information can be explained in 
the context of focus group discussions about 
groups. Although participants ranked group 
meetings as relatively unimportant, they 
were a large part of ICRAF's method. 
ICRAF worked with women's, church, 
and youth groups, and founded geographi- 
cal committees (called Sub-Location Area 
Committees [SLAC]). Each of these would 
periodically meet with ICRAF agents, and 
discuss grievances and techniques. Yet no- 
body mentioned that these groups dissemi- 
nated any information outside of the PRA 
exercises. Much to the contrary, both men's 
groups mentioned that group members "lead 
by example" — meaning, we believe, that 
they did not do very much direct dissemi- 
nation, but would adopt technologies and 
expect others to copy them. This helps ex- 
plain why participants rank local groups at 22 
percent. Similarly, every focus group com- 
plained that participating in groups had one 
negative side effect. Participation meant that 
you could not attend local funerals, or church 
services, since their meeting times frequently 
overlapped. Thus, funerals and church were 
not good sources of information. 

Nonetheless, it is a mistake to believe 
that local groups were entirely ineffective 
in accomplishing ICRAF's goals. Groups 
were responsible for both bringing issues 
to the disseminators' attention and for 
strengthening social ties. According to the 
farmers' focus group discussions, locals 
were instrumental at changing a number of 
dissemination practices. According to both 
men's focus groups, residents "demanded" 
that local committees bring in other tech- 
nologies, such as dairy cows. These two 

focus groups also reported that residents 
demanded that the SLAC send members to 
other villages to learn about other technolo- 
gies. Finally, all focus groups reported on 
demands to ICRAF that membership fees be 
waived or lowered (ICRAF had suggested 
that SLACs collect fees to provide for their 
operational expenses), and that meetings 
be held when no funerals or markets were 
underway. While the AIDS epidemic and 
importance of local marketing practices 
make it unlikely that the meetings could 
have accommodated everybody, people were 
grateful that local committees made efforts 
to correct this. In addition, all groups except 
non-poor men said that a great benefit of 
these ICRAF-inspired groups is that indi- 
viduals have become more confident as a 
result of membership. All focus groups men- 
tioned that the new groups had brought the 
community "closer together," which is per- 
haps their greatest accomplishment. It seems, 
then, that ICRAF has made progress toward 
realizing its goal of improving social capital 
by improving local groups. While this as- 
sessment is positive, areas for improvement 
emerged from the focus groups and PRA 
exercises. One way that disseminators appeal 
chiefly to the wealthy is that they dissemi- 
nate much of their information in Swahili, 
rather than local languages. Interestingly, 
it was non-poor men's and women's focus 
groups who complained about this (al- 
though they should be in a better position 
to understand than poor farmers), and noted 
that it limited understanding. In addition, 
both women's focus groups reported that 
they felt intimidated by men at groups, and 
that this limited their ability to learn. This 
was expressed with exceptional strength 
when women reported that some men had 
stolen group funds. Further, all groups ex- 
cept non-poor women reported that local 
elites were stronger as a result of these 
groups. This is one possible reason why 
poor men reported that they gain less than 
men of similar status elsewhere. They are 
cut out of elite circles. Similarly, even less 
poor men are largely not drawn from local 


social elites, and therefore do not gain as 
much information as they could. In contrast, 
women started with lower levels of knowl- 
edge in the first place. Thus, they gain more 
knowledge about technologies. By making 
these groups more inclusive of non-elite 
people, disseminating organizations could 
also make local groups more responsive to 
other local needs. For instance, all groups 
reported that the way they were taught to 
use tithonia was too hard for old people 
to use, since it is too labor intensive. Groups 
did not address these concerns. 

ticipants are most positive about this ap- 
proach, as it allows them to shape local pro- 
grams through their participation. None- 
theless, participants have some complaints 
about their participation level. Notably, they 
complain that the approach does not include 
everybody. These complaints are illustrated 
when participants compare MoARD to an 
organization of superior resources, the 
United States' Centers for Disease Control 
and Prevention (CDC). These offer some 
ways for MoARD to improve dissemination 
of SFR technology. 


Making these groups accessible to non- 
elites would also make them more effec- 
tive at disseminating information. Although 
ICRAF hoped to use local groups, these 
groups have drawbacks. Instead, farmers 
reported that most of their social learning 
occurred via formal means as explained ear- 
lier. It should come as no great surprise that 
groups dominated by elites were not espe- 
cially effective at disseminating information 
to the majority. 

Disseminators should be cognizant of 
this potential for domination and revise ap- 
proaches in a way that could reduce this in- 
fluence, so that they do not become agents 
of social stratification as well as technology. 
Perhaps much of this fault rests with 
ICRAF's assumption that residents are part 
of groups outside kinship. Some people 
obviously are, but others are not. ICRAF 
should redouble their efforts to reach those 
people outside groups, as people reported 
that they learned more about SFR outside 
groups. In spite of this, people in Sauri are 
gaining much from ICRAF's interventions. 

The Catchment 
Approach in Gongo 

Congo represents an interesting case for a 
variety of reasons. For one, it was chosen as 
a research site that represented the catch- 
ment approach as used by MoARD. Also, 
organizations used a variety of methods. Par- 

Assessment of 
Disseminating Organizations 

Congo had contact with more external or- 
ganizations than most other villages. Partic- 
ipants readily identified CARE, KARI, and 
MoARD as locally active. Wealthier men 
and women also reported that ICRAF 
worked in the community, although they 
provided little other information on this. 
In addition, Gongo saw many other groups 
working there. A group called Africa NOW 
attempted to set up irrigation and water sys- 
tems. One called ICA attempted to teach an- 
other technique for biomass transfer. Kenya 
Seed set up subsidized seed buying pro- 
grams. Several hospitals ran public health 
education programs. Residents could con- 
ceivably have learned about everything from 
sunflower raising, to compost, to credit 
from groups that ranged from the local reli- 
gious organizations, to a World Bank team, 
to the CDC. 

Although focus group participants some- 
times had difficulty associating organizations 
with specific technologies or methods, they 
generally had a positive association with 
the organization they closely associated 
with the community. Their assessments of 
MoARD do particularly well. Gongo partic- 
ipants ranked MoARD as more useful than 
participants in other communities, giving it 
an average score of 50 percent. Poor men 
and women ranked MoARD particularly 
high. Poor men scored it at 75 percent, while 
poor women gave it 62 percent. While the 


assessment goes down among non-poor men 
and women (to 27 and 33 percent, respec- 
tively), every number is higher than the 
average ranking of MoARD at 23 percent, 
and 21 percent in Luo communities. 

Other agencies did not do as well. Al- 
though it still scores higher than the average, 
KARI scores lower at 35 percent. Of course, 
this is still much higher than its Luo average 
of 15 percent, and this gain is significant 
since KARFs involvement in the commu- 
nity was limited. KARI did better among 
non-poor men and women, both of whom 
ranked it at 45 percent. Local assessments 
of CARE indicate this difference between 
non-poor and poor more dramatically. In 
this case, poor men and women each ranked 
CARE at about 5 percent. In contrast, non- 
poor villagers gave it 25 percent. Only 
non-poor men and women identify ICRAF 
as working in the community, and nobody 
attributed any dissemination to it. 

Some of these differences can be ex- 
plained by histories of organizations in 
the community. Few participants identified 
ICRAF as working in the community at all. 
This is probably because the organization 
was new to the area, and had contacted local 
leaders and organizations only to this date. 
It is less obvious why CARE and KARI 
would rank lower, but reasons are suggested 
by the role of local groups. Both organiza- 
tions work mostly with local groups. Espe- 
cially in the case of KARI, focus group 
participants drew strong links between the 
organization and local committees, most of 
which existed before the programs began. 
These committees were predominantly com- 
posed of wealthier residents. An example of 
this comes from a story told by both poor 
men and women in the community. KARI 
wanted to create a model farm in Congo, of 
a farm size larger than that desired by or to 
be found among poorer farmers. Both poor 
groups commented on the exclusion they 
felt. This ended up further weakening par- 
ticipants' associations with KARI, and 
KARI opened up the project to all who 
wanted to participate. Although no similar 

story exists for CARE, participants reported 
that they also work with committees and that 
even their work with schools limited the 
poor's participation, since school attendance 
is not universal, and the school clubs they 
used to disseminate SFR were better at- 
tended by wealthier children. 

Teaching Methods and 
Knowledge Gains 

MoARD's positive assessment stems from 
its good work that participants saw as posi- 
tive and participatory. In Congo, MoARD 
entered the community via assistant chiefs' 
barazas, and moved on to encourage par- 
ticipation among all residents. They partic- 
ularly focused on active participation via 
local committees chosen by elections in dif- 
ferent clans. These committees would then 
receive training on how best to disseminate 
agricultural technology, including those 
related to SFR. From there, committees per- 
formed a variety of dissemination methods. 
These ranged from tours of model farms, 
to general meetings, to barazas. They also 
were expected to visit individual farms to 
encourage use of SFR. Residents were sup- 
posed to learn a variety of techniques of soil 
fertility improvement, soil and water con- 
servation, farming practices, and accounting 
practices such as double entry bookkeeping. 
Finally, MoARD attempted to follow up on 
these techniques with monitoring and eval- 
uation that would allow them to identify their 
shortcomings and to generate new areas for 
improvement. The result of this was that par- 
ticipants reported back good assessments 
of the technologies, methods, and staff of 

Participants were generally positive 
about the effects of the technologies 
MoARD sought to introduce. Poor farmers 
were particularly positive about the results. 
Participants saw their improved yields as 
the direct result of the technologies. Al- 
though knowledge gain on most technolo- 
gies was substantial, among the eight tech- 
nologies participants learned about (tithonia, 
crop residue, farmyard manure, compost, 


commercial fertilizer, rock phosphate, im- 
proved fallows, and terraces), they reported 
learning more than the average across the 
six villages only on crop residue and com- 
mercial fertilizer. These scores are some- 
what biased by non-poor participants, as 
poor men did not mention that they gained 
any knowledge about crop residue and poor 
women reported only a 40 percent knowl- 
edge gain, compared to 63 percent for all of 
Gongo. Similarly, poor participants learned 
slightly less than others about commercial 
fertilizer. Poor participants reported a 47 per- 
cent knowledge gain. Non-poor partici- 
pants reported a 60 percent knowledge gain. 
Participants reported that they learned less 
about five technologies (tithonia, farmyard 
manure, compost, rock phosphate, and im- 
proved fallows) than participants in other 
villages. On tithonia, rock phosphate, and 
improved fallows, with average knowledge 
gains of 14 percent, 4 percent, and 7 per- 
cent, respectively, knowledge gains are quite 
low. Nonetheless, all groups commented 
on the value of these technologies. The fact 
that the poor were especially positive on 
the utility of the technologies they learned 
from MoARD indicates that, although they 
do not believe they learned as much as 
farmers in other communities, the knowl- 
edge they gained was especially applicable 
and beneficial. 

This assessment suggests that SFR dis- 
seminators in Gongo are using effective 
teaching methods. As in most communities, 
people prefer one or two methods. The chief 
difference in Gongo is that participants 
prefer methods less popular elsewhere. But, 
as in most places, assessment of methods 
varies by wealth status. Here, the most pop- 
ular method among the non-poor was group 
meetings. Among the poor, the most popu- 
lar method was observation. What these two 
methods share is that they both fundamen- 
tally involve inclusion. Rather than present 
model versions of technology removed from 
much of the social and ecological context 
where farmers are likely to use them, these 
methods allow residents to see technology 

in action, participate in its development, 
and to ask questions freely as they see fit. 
This is what all focus groups said they liked 
about the methods. Local participation led 
MoARD to emphasize dairy farming, nutri- 
tion, and canning among the technologies 
they disseminate. Formal presentations, such 
as demonstrations, tours, training, and com- 
mittee meetings, score far lower in Gongo 
than in the rest of the communities. 

The greatest challenge to these methods 
arises from the preference of different social 
groups for different dissemination tech- 
niques. Although gender preferences do not 
differ from the overall pattern, with women 
preferring barazas, meetings, and tours, 
wealth status preferences differ greatly. As 
noted earlier, the non-poor group prefers 
meetings and the poor prefer observations. 
Group meetings are specific social contexts, 
whereas observations can occur any time 
and any place. This indicates that, although 
it generally succeeds, MoARD has not di- 
rectly reached poor residents as much as it 
would like. 

Local Disseminators 

It is difficult to distill a pattern on attitudes 
toward local groups. Only poor men and 
non-poor women mention the catchment 
committees, giving them scores of about 1 1 
percent. "Other local groups" are given a 
score of 39 percent by poor men, but less 
than 10 percent by all other groups. Poor 
men also favor learning from "other farm- 
ers," giving them a score of 20 percent, over 
twice as high as that from any of the other 
three groups. In contrast, churches are seen 
as useful forums by all groups except poor 
men. Barazas receive the most consistently 
high scores of between 20 and 30 percent, 
but poor men do not mention them at all. 
Trying to understand the wide variations in 
this scoring would take additional research. 
Additional information about people's ex- 
perience with groups is, however, provided 
by the mapping of village information flows, 
where poor men and women showed either 
no link or a weak link between MoARD and 


local groups or farmers. In contrast, non-poor 
men and women both said that MoARD 
has strong two-way interactions with groups 
and farmers. This appears somewhat con- 
tradictory, given the high scores that poor 
groups gave MoARD. This may indicate, 
however, that people recognize the impor- 
tance of MoARD in providing information 
by different means, but that poor farmers 
had less experience in groups on which to 

Participants' reports on local social cap- 
ital further verify this. In their discussions, 
non-poor men and women both reported 
that the presence of the groups in the com- 
munities strengthened local solidarity. And, 
from their perspective, they may have. The 
fieldworker researching Gongo reported that 
in practice, the means of choosing groups 
meant that clan elders were selected. In 
societies with clans, an elder is necessarily 
both older and from the dominant lineage. 
In other words, those serving in the groups 
were the oldest, wealthiest, and most pow- 
erful residents. It is understandable then 
that both poor men and women reported 
that local groups were chosen unfairly, and 
wanted this rectified in the future. To their 
credit, non-poor men and women both ad- 
mitted that group members get benefits first, 
and that they give these benefits to their 
close kin before others. 


It should be noted, however, that although 
lack of participation is a problem that should 
be corrected, participants are generally pos- 
itive about MoARD and the information 
they bring. But it should also be noted that 
MoARD is not the most highly regarded 
group disseminating technology in the area. 
That honor is held by the CDC. Considering 
what the CDC does can also offer valuable 
lessons for those seeking to distribute SFR. 
Participants presented almost a constant 
stream of praise about the organization, be- 
cause the CDC provided a technology that 
people saw as extremely useful and well dis- 
tributed. They provided antimalaria technol- 

ogy, specifically nets that keep mosquitoes 
from breeding, and treatment information for 
malaria sufferers. The community clearly 
desired these benefits. All focus groups 
reported that malaria was a problem in the 
community. Furthermore, all focus groups 
reported that the technologies worked and 
were easy to use. To top it off, the technolo- 
gies were given to all residents for free. The 
CDC also held well-attended seminars and 
had their staff regularly visit residents to 
ensure that they were following guidelines. 
It is perhaps unfair to compare other 
organizations to the CDC. No other group 
could possibly have provided such direct and 
clear benefits as malaria prevention. Nor 
could any other group marshal the trained 
staff used by the CDC. But, although the 
comparison may be unfair, it also provides a 
valuable lesson to other groups. While SFR 
is not as beneficial as prevention of an en- 
demic, chronic, and debilitating disease that 
can strike anybody, it is important to note 
that the CDC did other things well. Al- 
though agricultural disseminators cannot 
purchase inputs for farmers, which partici- 
pants are in favor of, they could do a better 
job of trying to reach all community mem- 
bers with their interventions, as did the 
CDC. Similarly, by all accounts, the CDC 
patiently explained the benefits of their 
intervention. They also listened to local 
concerns and answered questions. This pro- 
vides ideas for how MoARD and other SFR 
disseminators can be more effective in fur- 
ther efforts. 

The Trace Approach 
in Muhanda-Arude 

Muhanda-Arude is a Luo village located 
in the Siaya District. CARE began working 
with tree management and tree nurseries, 
more than SFR technologies, promoting, for 
example, Leucaena and calliandra. This 
work began in the mid-1980s and CARE 
stayed until 1999, although farmers said that 
the real intervention ended earlier. ICRAF 
picked up SFR activities in the village later. 


Some villagers recall work on soil fertility 
done by MoARD as far back as the 1970s, 
though there is little recollection of this and 
it is not evaluated here. Muhanda was cho- 
sen as a site for the dissemination focus 
groups because (1) it was one of the earlier 
villages to receive the technology, and thus 
there is a greater time horizon for evaluating 
the technology; (2) it was one of the villages 
where the survey and case study work was 
done for this project, allowing some overlap 
in site selection; and (3) CARE used its 
TRACE approach there (Training Resource 
Person in Agriculture or Agroforestry for 
Community Extension), an important form 
of intervention that was used throughout 
Siaya. As one of our stakeholders, CARE 
has an interest in the results of the study. 

As described in the main report, 
TRACE was an effort to promote sustain- 
ability through a network of farmers, local 
groups, and regional groups that would con- 
duct needs assessment and promote training. 
Officially, the structure involved regional 
groups called Locational Agroforestry Com- 
mittees (LAC, later changed to LASCO), 
which were to select groups, which, in turn, 
chose Group Resource Persons and several 
adaptive research farmers. In Muhanda- 
Arude, however, these structures were not 
operating as envisioned, at least in the per- 
ception of farmers. Still, some effective 
training took place and people are using 
the technologies. 

Teaching Methods 

As in the other villages, different types of 
training forums were used by these organi- 
zations. Among the focus group participants 
as a whole, barazas and "general meetings" 
were by far the most popular, especially 
among poor women. This suggests that 
people in Muhanda-Arude are most com- 
fortable with forums that involve the whole 
community, rather than sessions that in- 
volved a more select group of farmers, such 
as tours and exchange visits, which were 
the least popular. Both poor and better-off 

women explained the importance of the 
village elder in bringing farmers together to 
inform everyone of the presence of one of 
these external organizations in the village. 
Although some say this system works, some 
poor women said they never receive this in- 
formation, and suggest that the village elder 
invite every household in the village directly. 
One of the stronger messages from the 
research was that all four groups — men, 
women, poor, and better-off farmers — 
expressed that the community at large needs 
to be more involved in processes of tech- 
nology dissemination than it has been in 
these past interventions, rather than the com- 
munication being primarily with a few. 

Observations of other farmers' fields 
were also a popular method of learning 
compared to most other methods, somewhat 
more so in Arude than in four of the five 
other villages, though this was only among 
rich men and women. Of these, the women 
preferred this method far more than men, 
which is consistent with the hypothesis that 
women benefit more from methods that are 
inclusive and open to everyone, rather than 
a smaller invited group. Demonstrations, 
field days, meetings (of smaller groups, as 
opposed to "general meetings"), oral con- 
versations, "training," and exchange visits 
had a low popularity among both men and 
women. Neither women's focus groups 
mentioned field days, exchange visits, or 
small meetings, suggesting that women par- 
ticipated less in these. The most popular 
methods for poor farmers were the "general 
meeting" and baraza, whereas rich farmers 
do not mention general meetings at all. In 
contrast, only rich farmers mentioned ob- 
servations, oral conversation, and exchange 
visits. Most interesting is that none of the 
four groups mentioned funerals; Muhanda 
is the only village among the six where fu- 
nerals are not mentioned at all in the focus 

On average across the four groups, 
CARE was given a rating of 60 percent 
compared to ICRAF's 40 percent, indicat- 


ing that both organizations were popular 
but CARE more so. Possibly because of 
CARE's longer-term involvement versus 
ICRAF's more recent intervention, CARE 
is remembered for having done more in 
the community. However, CARE is viewed 
much more favorably by the better-off 
groups than by the poor groups, which favor 
ICRAF substantially. In spite of this scor- 
ing, poor men remarked that CARE worked 
for a long time in the village and benefited 
the farmers who worked closely with the 
organization, and poor women appreciated 
the gardening tools that CARE left for use 
in the nursery. Men and women view the 
two organizations essentially equally, con- 
sistent with both focus group comments and 
survey results, indicating that disseminating 
institutions were not discriminating by 
gender. However, when asked about the 
strength of communication between the or- 
ganizations and farmers, only poor women 
said communication was strong with both 
CARE and ICRAF, with the other groups 
saying communication was weak or non- 
existent with both organizations. Only the 
poor men remember Mo ARE), saying that 
there was strong two-way communication 
with MoARD at the time, although the over- 
all assessment of MoARD was lower than 
for CARE and ICRAF. No women mention 
MoARD, possibly because in the early 
1970s, the government was not paying much 
attention to women as farmers. 

Poor and better-off women mentioned 
that the venue for meetings needed to be 
more suitable for participation, probably 
reflecting their greater restrictions on travel- 
ing. The better-off women pointed out that 
these far away venues are used simultane- 
ously for trading, leaving insufficient time 
for the training, and that they would prefer 
organizations to go from home to home 
visiting farmers. 

One dissemination method of particular 
concern to farmers is the "contact farmer" 
or "adaptive research farmer" approach, 
whereby the external organization works 

primarily on the farm of individual farmers, 
who, in turn, are expected to serve as 
demonstrators to other farmers in the vil- 
lage. Villagers reported that at first CARE 
worked with individual farmers, and then 
changed to a group-based approach. How- 
ever, the TRACE approach still uses Adap- 
tive Research Farmers (ARFs). ICRAF used 
these adaptive research farmers as well. 
This approach has caused problems, and is 
disliked by both men and women and by rich 
and poor groups alike. According to partici- 
pants, these contact farmers are resented 
because they are seen as (1) unfairly fa- 
vored and (2) do not spread their knowledge 
to others. Both groups of men in particular 
stressed that the contact or adaptive re- 
search farmers do not share information, 
and according to the better-off men, they are 
seen as "the wealthy and educated who are 
frequently visited and make others feel left 
out and different from the preferred farm- 
ers." The constant attention from govern- 
ment and other institutions is said to be a 
problem for the rest of the village, as well 
as the contact farmers themselves in terms 
of time it takes up. Non-poor men also said 
that contact farmers were picked by the in- 
stitutions, and better-off men commented 
that most people did not even know of the 
designations. This is contrary to the intended 
position of CARE and ICRAF that farmers 
themselves should choose the contact farmer. 
Poor women criticized the contact farmer for 
not sharing seeds, saying that people waste 
their time trying to get things from them. 
However, they saw one CARE contact 
farmer in a positive light, recalling that he 
helped with training and, more significantly, 
brought women together to form women's 
groups to promote the technology. 

Notwithstanding local perceptions, it is 
important to recognize the role of the con- 
tact or adaptive research farmer from the 
perspective of the disseminating organiza- 
tion. As CARE points out in its TRACE 
approach, technologies that work in one re- 
gion may not be adapted to a new region. 


Adaptive research farmers are crucial for 
testing technologies and practices and 
adapting them to local conditions, before 
they are disseminated to other farmers. 
There may thus be a period in which there 
is considerable contact between the ARF 
and the external organization, before 
many other farmers are brought into the 
process. It is equally important, however, 
to recognize that the social context will 
affect outcomes, and that this method as 
currently practiced is problematic in the 
context of local social relationships, and has 
significantly affected the way in which peo- 
ple respond to the external organizations as 
they introduce technologies. It is necessary 
then to bring the community more widely 
into the learning process at an earlier stage, 
to make sure people understand the role of 
the ARF and approve of the choice. 

Evaluation of Training 

TRACE involves a training needs assess- 
ment, in which the groups chosen by LAC 
are consulted regarding what kind of train- 
ing is needed in the village. Focus group 
participants were not aware of this process 
(it is possible that few if any were part of 
this consulted group). People found the train- 
ing and the technologies useful and effec- 
tive, however. SFR technologies that people 
learned about include tithonia, crop residue, 
farmyard manure, compost, commercial fer- 
tilizer, rock phosphate, improved fallow, 
and terraces. Additional training took place 
in tree planting, nursery management, im- 
proved cookers, and kitchen gardens. The 
"ladders" exercise where participants rated 
the percent increase in their knowledge of 
the SFR technologies suggested consider- 
able learning took place during the inter- 
vention and that knowledge has continued 
to increase through practice for some of the 
technologies. In the discussions, however, 
men were critical of the depth of the training 
and the number of people who benefited. 
Some men said that farmers were using 
these technologies but with a "shallow idea" 

and that inadequate training and demonstra- 
tion was a constraint on uptake and the ben- 
efits that flow from adoption. Rich men said 
that CARE did serious training, but that 
only a few benefited from the training, 
particularly the group members (see later), a 
comment they made about ICRAF as well. 

Women spoke more favorably about the 
training in several technologies than did men, 
with rich and poor women emphasizing the 
training that CARE gave in tree planting and 
that women were planting trees and benefit- 
ing from them. ICRAF was also said to have 
provided useful training. In the ladders 
exercise, poor farmers report they have 
learned more than rich farmers on the more 
easily available technologies such as ter- 
races, tithonia, composting, and improved 
fallows, with the greatest knowledge growth 
in the first two technologies. Richer farmers 
learned more about commercial fertilizer 
and rock phosphate (knowledge gain was 
the lowest on rock phosphate for all four 
groups). Of all groups, poor men stand out 
as having learned the most about tithonia 
(80 percent knowledge gain), terraces (80 
percent), and improved fallows (70 percent). 
Women tend to feel that they learned more 
than men on all technologies (those men- 
tioned thus far as well as farmyard manure 
and compost), except improved fallows, pos- 
sibly because their starting point of knowl- 
edge and training is generally lower than 
men's. The rich women's group in particular 
talked at length about all they had learned 
through CARE, ICRAF, and MoARD. This 
is significant because it suggests that women 
are benefiting from the interventions, at least 
in terms of human capital development. 

CARE's program of training through the 
schools is viewed as successful in Arude. 
All four groups said that children are learn- 
ing about tree planting and agriculture. 
Children plant vegetables and trees at home, 
and parents are said to learn by observing 
this planting. Women say that sale of veg- 
etables at home produces income for the 
family, and men point out that the trees 


have produced income and timber for the 
schools, and that the trees are still there, 
although rich men said that there are fewer 
than before. No problems were raised with 
the schools program. 

Group-Based Extension 

CARE started out using contact farmers, but 
changed to a group-based approach when 
the former method was found not to be work- 
ing well. Both women and men seemed to 
remember the women's groups formed by 
CARE more than men's groups, and the for- 
mer were remembered as having been suc- 
cessful for a time. One of these groups had 
as many as 40 members. The men's groups 
seem more informal, without identified 
names (as the women's groups). This seems 
to be consistent with greater group-based 
activity in general among women than men. 
Men said that few farmers were in- 
volved in these groups, and that while they 
benefited their members, this was only a 
small group and most villagers have not 
benefited. This is confirmed by the Venn 
diagram by rich men that shows a link be- 
tween CARE and the contact farmers, but 
not the groups, and no link between the 
group and other farmers, or between the con- 
tact farmer and the groups or other farmers. 
There is a strong link between ICRAF and 
the groups, however. Rich men said that the 
committee organized field days and gave 
demonstrations to farmers, and one group 
member offered his land for nursery man- 
agement and training. They said that CARE 
did more of the organizing of farmers than 
farmers themselves, however. In the ranking 
exercise for internal organizations, only the 
group of rich men mentioned a committee 
and ranked it lower than the other organiza- 
tions. Significantly, the groups seem to be 
working better for rich men than for poor 
men, evidenced by the fact that poor men 
did not mention groups in the ranking or in 
the Venn diagram. One issue related to sus- 
tainability of the methods is whether groups 
or farmers are willing to pay for extension 

services. Although there is no evidence of 
this having occurred with SFR, rich men 
said they would be willing to share costs 
for a technology if it would be beneficial 
to them. 

Other internal sources of information 
were barazas, other farmers, schools, 
women's and church groups, and local 
leaders/elders. Rich men ranked local leaders/ 
elders and other farmers highest, while poor 
men only mentioned the first three and pre- 
ferred the barazas. Women viewed barazas 
less favorably than men, and poor women 
ranked other farmers, women's groups, and 
churches highest. Curiously, rich women did 
not mention women's and church groups in 
the ranking. 

The CARE contact farmer assisted 
women in Arude in setting up some new 
women's groups for the purpose of pro- 
moting the trees, and poor and rich women 
said they benefited from these groups. In 
the Venn exercise, poor women showed that 
both CARE and ICRAF had strong links 
(although unidirectional) with the women's 
groups; however, rich women showed no 
links between the women's group and either 
organization. Furthermore, the women's 
groups collapsed after CARE left, suggest- 
ing the value of using existing group struc- 
tures, rather than establishing new ones for 
the purpose of a new intervention. It is 
likely that, especially under conditions of 
an AIDS crisis, groups are strained for re- 
sources and members strained for time, and 
that those with social insurance functions 
will be the strongest. Rich women, however, 
did say they would like to have women's 
groups dealing with agriculture, and women's 
church and welfare groups also serve as 
agriculture groups in some cases, for exam- 
ple, producing maize and beans for funerals. 
This could suggest the value of more agri- 
culture activities being channeled through 
existing groups. One small women's group 
involved with growing beans, maize, and po- 
tatoes continued its activities and hoped that 
ICRAF would train them in new farming 


techniques and SFR to increase their yields. 
The groups are seen as potentially providing 
benefits in addition to training, such as 
seeds and credit. 

In general, there do not seem to be many 
two-way information flows between exter- 
nal and internal organizations or individu- 
als. This implies that although farmers may 
be receiving good information, the system 
by which external organizations learn from 
the input of farmers does not seem to be 
working well. Poor and rich women showed 
a two-way link only between the chief 39 
and women's groups (rich women included 
friends and relatives). However, rich men 
do show a strong two-way link between 
ICRAF and the local groups, and poor 
men showed a strong two-way link between 
CARE and the contact farmer. 

The LAC and LASCO structures did not 
appear to be functioning in Arude, or were 
perhaps not visible widely enough to be 
known by the farmers in the focus groups. 
This was confirmed by rich men who said 
that LASCO is known only to a few farm- 
ers, that they have not been effective, and 
no longer exist. Poor men said that neither 
LAC nor LASCO existed in the village, and 
women did not mention them at all. Below 
are notes from the focus group with rich 

CARE, which did some serious training 
though only a few especially group 
members benefited. A majority in the 
village did not benefit. There are only 
two adaptive research farmers in the 
village but not known by this designa- 
tion. Most participants did not know of 
these designations or existence of any 

committee. Only one out of ten [focus 
group] participants knew of LASCO 
and was a member of the committee. 
The committee was active only during 
the time of CARE and could organize 
for field days on occasions to demon- 
strate to farmers what they need to do. 
Since the approach targeted group 
members only, they are the majority of 
beneficiaries. The committee (LASCO) 
as well as the groups collapsed after 
CARE; hence no activity is now done 
by committees. 

Poor men said that some of the extension 
activities brought the community together. 
The groups introduced or were affected by 
certain social tensions, however, some of 
which contributed to their collapse. The root 
of this, as described by the men's groups, is 
that some village members benefited from 
the groups while others did not: the greater 
attention that these farmers had from CARE, 
uneven distribution of resources, the ability 
of some to amass wealth through the pro- 
cess, and conflicts over resources — all fos- 
tered resentment and politics. There was also 
mismanagement of resources among both 
the men's and women's groups after CARE 
left the village, which, in turn, increased so- 
cial tensions within the village. Villagers felt 
that CARE left the village too soon, before 
the groups where able to stand on their own 
feet, and that there was a need for greater 
monitoring and follow-up. Part of the prob- 
lem may lie in the fact that CARE envi- 
sioned LACs to carry out monitoring and 
evaluation, and if the LACs were not func- 
tioning well for this community, this function 
could have slipped through the cracks. 

39 These village-level studies refer often to "chiefs" as involved with dissemination through their serving as liai- 
son between external organizations and farmers, holding barazas, and involvement with other aspects of dissem- 
ination. Technically a chief is the head of a location and the subchief is the head of a sub-location. Most of those 
involved in dissemination would be subchiefs because a chief has thousands of households under his jurisdiction. 
Given the many other obligations of a chief, his/her role as a disseminator would be limited, while assistant chiefs 
would be more directly involved. In the focus groups, however, participants usually referred to "chiefs" even if 
they usually meant assistant chiefs. We have thus chosen to use their language and keep the word chiefs. 



As indicated in the survey, the case studies, 
and the focus groups, the SFR technologies 
are seen as valuable and people welcome 
the training, and have learned a considerable 
amount about the technologies. However, the 
group- based dissemination methods them- 
selves do not appear to have functioned as 
envisioned, with lack of participation and 
sustainability over the long term. The expe- 
rience in Muhanda suggests that more time 
and effort would have been necessary for 
building capacity in groups, and that addi- 
tional monitoring after the dissemination 
intervention ended could have helped to 
identify problems and resolve them where 
possible. To a certain extent, the tensions that 
arose around the groups, as well as those 
around the contact farmers, are, however, 
social dynamics that can never be entirely 

The PLAR Approach 
in Mutsulio 

All of the disseminators covered in this 
study attempted to include the opinions of 
community members in their methods. In 
some cases this attempt aims to improve dis- 
semination methods or to point out defi- 
ciencies in the approach. The Participatory 
Learning and Action Research (PLAR) ap- 
proach in Mutsulio represents an attempt 
to go further. Here, local people ideally not 
only served as checks on the disseminating 
organisms, but also participated in develop- 
ing the programs. The program met with 
success on many levels. It achieved im- 
pressive results in knowledge gain. It also 
achieved impressive dialogue with local 
groups and local farmers. The approach also 
led to a heavy dependence on local groups 
and meetings to the exclusion of other 
methods and internal organizations. The most 
notable problem with this dependence was 
that more than other methods, this benefited 
some to the exclusion of others, and has 
caused enthusiasm for it to wane over time. 
The initial focus of PLAR was on soil fer- 

tility. Households were ranked according 
to better or worse soil managers and then 
groups were formed around similar manage- 
rial types. The focus of the PLAR approach 
was thus greater than agroforestry, but less 
than the range of technologies disseminated 
by the MoARD extension program. 

Assessment of 
Disseminating Organizations 

As noted in greater detail in the methods 
section, the approach used to disseminate 
SFR technology in Mutsulio was specifi- 
cally designed to increase local participation 
at all levels. In this it is commendable, as a 
frequent complaint residents voiced in other 
communities was that they were not suffi- 
ciently involved in planning and improving 
SFR projects. Although eliciting local opin- 
ions may sound simple, this approach shows 
it is not so. This appears to be one of the 
most elaborate of the approaches in these 
communities, as well as one of the best 
planned dissemination approaches covered 
here. Fundamentally, the approach sought 
to aid dissemination through participation, 
but also by appealing to a number of moti- 
vations. It sought to help farmers learn new 
soil fertility methods through "on-farm learn- 
ing, self-discovery, and experimentation." 
Disseminators hoped to satisfy people's cu- 
riosity about technologies they saw their 
neighbors using, and to thereby empower 
farmers. They hoped that it would increase 
farmers' capacity to join groups, speak for 
themselves, and positively affect their own 
lives. The approach set up a means to in- 
clude farmer input at most levels. Although 
the specifics of the program are complex, 
the four "steps" of its approach all involved 
input and interaction of farmers and exten- 
sion staff. For instance, farmers' groups were 
supposed to draw up action plans for dis- 
semination into the community. Similarly, 
farmers would suggest the topics to be 
covered in a general baraza that would ex- 
pose the whole community to the technology. 
Key to the approach is that "open-minded" 
farmers are active in groups, which meet 


often so that agents can discuss technolo- 
gies. They are then urged to try out the tech- 
nology, see if they like it, and disseminate it 
to others. Ultimately, their goal was to get 
farmers to suggest technologies themselves, 
and to figure out ways to improve them. 

The implementation of this program 
looks different from this ideal. In practice, it 
is most notable for differences in degree of 
group activity. The disseminating organiza- 
tions that were active here, KARI, MoARD, 
and ICRAF (although ICRAF's was an in- 
termittent and supportive role to KARI), all 
entered the community in more or less the 
same way, and went on to found numerous 
village committees. It is not surprising that 
the organizations proceeded in more or less 
the same way, as MoARD introduced the 
other two into the community. In addition, 
all used MoARD's means of entering the 
community. All organizations entered and 
were introduced via the same local man, 
who introduced extension staff at a baraza. 
Once this meeting was staged, local groups 
were formed by electing members from other 
local groups, mostly clan, church, women's, 
and funeral savings groups. For this reason, 
local groups resembled those in other com- 
munities to a great extent. Like in other 
communities, participation in these varied 
by social status. Poor focus groups list an 
average of one group as existing prior to 
dissemination of SFR technology. Non-poor 
groups list an average of five. Most likely 
this indicates that wealthier residents were 
more active in local groups, and that, there- 
fore, groups formed by SFR disseminators 
were groups of the better-off residents. 

Importantly, outside organizations were 
supposed to be especially active in local dis- 
semination, especially in seeking feedback 
from villagers. There can be no doubt that 
they did this. Most notably, men's focus 
groups both said that KARI had either 
strong two-way or one-way interactions with 
farmers. The two women's focus groups, 
on the other hand, indicated that KARI had 
one-way interaction with farmers or groups. 
MoARD and ICRAF were both held to be 

active in the community by two focus 
groups. The most interesting figure from 
this is that ICRAF was held to be active 
only by the two men's groups. ICRAF thus 
appears to have not made as many efforts to 
reach women in this village. 

This information is reflected in focus 
groups' assessments of outside organiza- 
tions' usefulness and importance. As we may 
expect, KARI scores the highest on this as- 
sessment. The three focus groups that men- 
tion it rank it at slightly higher than 60 per- 
cent. In contrast, the three groups that say 
they get information from ICRAF rank it at 
slightly less than 20 percent. But the most 
surprising aspect of ranking concerns two 
facts. The first is that only one group, non- 
poor women, rank the three sources, even 
though all three were active. In contrast, 
poor men say that they obtain all of their in- 
formation from MoARD. Because nothing 
in either of the focus group discussions 
suggests animosity toward any disseminat- 
ing organization in particular, this diversity 
suggests something else; that is, that the 
focus group participants of Mutsulio gener- 
ally are unclear on which outside group is 
active in the community. 

Teaching Methods 

Focus group analyses of the methods used 
by SFR technology disseminators eliminate 
what little controversy exists about the enor- 
mous influence groups hold in this method. 
The most common method used in dis- 
seminating knowledge was group meetings. 
Three of the four groups rated this much 
higher in Mutsulio than in other commu- 
nities. Most notably, poor women ranked 
groups at 70 percent, compared to about 
35 percent in other groups. Similarly, both 
non-poor groups ranked group meetings at 
50 percent, five times the number reported 
elsewhere. Although poor men do not report 
on this method per se, they, too, use groups. 
Here, the most common method for receiv- 
ing information was "PLAR," which one 
participant reported meant going to groups 
and then trying out technology. The only 


other method remotely as popular was oral 
conversations. These, however, were re- 
ported only by the two non-poor groups. 
They ranked conversations at about 40 per- 
cent. But this, too, supports the preceding 
analysis in which relatively non-poor indi- 
viduals would be more likely to gain from 
conversations with group members. 

Such use of groups to the exclusion of 
all other methods has a number of problems 
that are not unique to Mutsulio. Like every- 
where, solely using groups for dissemination 
leads to members gaining disproportionate 
amounts of information. Perhaps the best il- 
lustration of this comes from a statement 
made by the fieldworker who did the re- 
search here. He reports that poor men "en- 
vision commitment and hard work as ways 
to spread technology, so that other farmers 
can observe the technologies as practiced 
by the committee members." According 
to notes, these farmers see no other way that 
committees disseminate technology. But 
diffusion takes time. The only people who 
could be assured of benefiting from these 
technologies were those to whom it was 
introduced in the first place. 

These groups also suffered from a num- 
ber of other problems. Most notable is that 
farmers do not participate in them. As ar- 
gued earlier, non-poor people are the most 
typical group members, meaning that the 
majority of people in these communities are 
not participants. Poor men note that they feel 
a great constraint to participating in groups 
because they do not have enough land to be- 
come group members. Although this may 
not matter as several also noted that they did 
not have enough land to adopt some of the 
technologies anyway, it was a barrier to 
participation. Similarly, women see groups 
as being forums for men. Others see them 
as dominated by men, if only passively. 
Non-poor women even expressed their 
reluctance to participate because of men's 
dominance. In particular, they noted that 
men sit through meetings and then reject the 
technologies when they get home without 
having said a word at the meetings. Others 

would go solely to reap the monetary remit- 
tances that some groups paid, which they 
would pocket without adopting the tech- 
nology. But the groups and disseminating 
organizations were dissatisfying for another 
reason. Three of the four focus groups re- 
ported that they did not talk to farmers as 
much as they said they would, and would 
simply ignore their requests when they did. 

These groups also suffered from a num- 
ber of logistical and technical problems. All 
four focus groups reported that they are less 
likely to attend meetings or to be active in 
groups owing to their overlap with funerals. 
Similarly, poor women noted that they in- 
terfered with domestic chores, as the meet- 
ings could stretch for four hours. Both poor 
and non-poor men noted that organizations 
had failed to deliver on their promises sev- 
eral times. These same two focus groups 
also suggested that some of the committee 
members were chosen poorly. In particular, 
they were not especially hard working, 
which caused dissemination to take place at 
a very slow pace. These problems are aggra- 
vated, as few other dissemination methods 
were in use. 

The ultimate test of this method's effec- 
tiveness, however, lies not only in comments 
on process and disseminating structures. It 
also lies significantly in the amount of in- 
formation people learned from it. In only 
one case did farmers say they gained little 
information: neither poor men nor non- 
poor women report that they learned any- 
thing about crop residue management. This 
is somewhat counteracted by the massive 
gain that occurred in non-poor men (70 per- 
cent knowledge gain in Mutsulio versus 
40 percent in other villages). In addition, it 
should be noted that poor men and non-poor 
women reported relatively small gains in 
crop residue management of 10 and 20 per- 
cent in all the study communities. Farmers 
report about the same level of knowledge 
gain on average in three of the four tech- 
nologies (tithonia, 60 percent; farmyard ma- 
nure, 46 percent; and compost, 55 percent). 
But the gains in commercial fertilizer, rock 


phosphate, improved fallows, and terraces 
are impressive. For example, poor women 
report a 63 percent knowledge gain in rock 
phosphate. In other communities, they re- 
port a 29 percent gain. Similarly, poor men 
report a 68 percent knowledge gain in com- 
mercial fertilizer, which compares favor- 
ably to the 33 percent reported elsewhere. 
All groups report at least one substantial 
gain over the average. It is impressive that 
these gains could have occurred over such a 
range of technologies, particularly as none 
was specifically targeted. 

Local Dissemination 

Focus groups have surprisingly little dis- 
agreement on their sources of information 
inside the community. Far and away the 
most commonly listed source of informa- 
tion here was groups inside the community. 
In fact, residents ranked this as highly as all 
the other sources of information combined, 
at just over 50 percent. Perhaps the most 
impressive number to emerge from the 
PRA exercise on internal forms of dissemi- 
nation was the assessment of local groups 
by poor women at 80 percent. The compa- 
rable figure for the study villages as a whole 
was 22 percent. 

The second highest figure to emerge 
from this PRA exercise was farmers' assess- 
ments of the usefulness and importance of 
other farmers as a source of information. 
Here, both poor groups ranked this about 
the same as the average. In contrast, the two 
non-poor groups ranked it about twice as 
important. Non-poor men rank it at 30 per- 
cent and non-poor women at 25 percent. 
These figures suggest that less poor people 
are the prime beneficiaries of this approach. 
Not only are groups most likely composed 
of the non-poor, but also the non-poor are 
more likely to know others of their kind. 
Therefore, they are more likely to know 
group members who can teach them more. 
As such, the two non-poor groups get more 
information from other farmers. 

A number of other differences appear on 
this list related to low rankings. Although 

barazas score slightly higher than average 
among poor men in Mutsulio (33 percent 
versus 31 percent), every other score is lower 
for all technologies. Notably absent from this 
list are funerals, schools, women's groups, 
church groups, and local leaders. In fact, 
based on our focus groups, one would con- 
clude that these are not among SFR dissem- 
inators. At the very least, they are not used 
with great frequency. 


Clearly PLAR can lead to impressive gains 
in knowledge. It may be less able to sustain 
this growth, however, owing to resentment 
created in the community and inability to 
truly engage different social groups, which 
undermines the goals of PLAR. It may have 
been more effective had they also tried other 
methods, or taken steps to try to ensure that 
the groups were not the domain of relatively 
few community members. Finally, this al- 
most complete reliance on groups in prac- 
tice leads to the inability to hear community 
concerns. Improving this method can allow 
disseminators to accomplish their other 
goals of improving local groups and leading 
to direct community involvement with tech- 
nology and empowerment. 

The Catchment 
Approach in Mwitubi 

Mwitubi village is in Mwitubi sub-location 
in a Luhya area of western Kenya. The 
catchment area forms the main focal point 
in the dissemination approach utilized in the 
village and it was introduced by Mo ARE). 
However, various NGOs have since entered 
and worked in the village, using a variety of 
other approaches to reach farmers with a 
variety of messages. The catchment ap- 
proach focused on a group of villages with 
the same water catchment and drainage sys- 
tem to tackle the issue of water and soil loss. 
MoARD organized farmers in this group of 
villages, covering the catchment into what 
has come to be known as a Catchment Com- 
mittee. The committee mobilized farmers 


to be trained and to provide labor in the 
soil and water conservation works within 
the catchment. Later MoARD also formed 
village committees in the catchment and it 
is the village committee that participants 
zeroed in on during their discussions, as it 
was fresh in their minds. 

MoARD worked with a group of vil- 
lages with common geographical orientation 
in order to deal with not just a common 
problem, but also through measures that 
complemented each other to deal with the 
problem in totality. The topography of the 
catchment was taken as presenting the total 
problem and conservation work in any one 
farm and village added up to solve the prob- 
lem. As work proceeded, the entire catch- 
ment was covered with conservation work 
and the whole problem of water and soil loss 
in the catchment lessened significantly. This 
approach was a deviation from earlier ap- 
proaches that focused on single farm units 
and failed because the problem was not 
tackled in the farms upstream and in those 
downstream. In this new approach, the 
problem of soil and water loss was taken as 
a community-wide (catchment) problem and 
community physical and human resources 
were mobilized to deal with it. The approach 
focused on providing skills to the people in 
the catchment to use local institutions and 
resources to tackle the problem at hand. 

Farmers in the village made terraces on 
individual farms before MoARD introduced 
concerted soil and water conservation mea- 
sures through the catchment approach. A 
catchment committee elected by the local 
farmers coordinated the activities in the 
catchment. Although training by MoARD 
focused on soil and water conservation, other 
agricultural matters were covered as well to 
help improve agricultural standards in the 

Farmers' Comparative Evaluation 
of the External Organizations 

When focus group discussions considered 
the organizations that worked with farmers 
in Mwitubi village, poor and non-poor male 

participants named MoARD, ICRAF, and 
FSDA, while poor and non-poor female par- 
ticipants named MoARD and ICRAF. The 
fact that female participants, both poor and 
non-poor, were unaware that FSDA had 
worked in the village may imply that women 
have limited access to information about the 
goings-on in the village unless they are tar- 
geted directly, as has been done by MoARD, 
for instance. FSDA did not undertake wide- 
spread dissemination activities in the village. 

In ranking the various organizations 
that had undertaken dissemination in the 
village, the poor and non-poor men's groups 
both gave ICRAF an 82 percent rating and 
MoARD an 18 percent rating. Poor women 
gave ICRAF 47 percent and MoARD 53 
percent, while non-poor women ranked them 
conversely, at 55 percent and 45 percent, re- 
spectively. Overall, participants rated ICRAF 
significantly higher than MoARD, in spite 
of the fact that the latter had worked in the 
village for a longer time. Part of the reason 
for this scenario was that farmers' interac- 
tion with ICRAF was recent and it provided 
transport for MoARD staff and, hence, was 
assumed to have replaced the latter. How- 
ever, non-poor women participants noted 
that the interaction between MoARD and 
ICRAF was smooth, with the latter using 
staff from the former to reach farmers in 
the village. The ratings did not vary signifi- 
cantly on the bases of the socioeconomic 
status of the participants, although women 
tend to view MoARD more favorably than 
do men, and poor women had a slight pref- 
erence for MoARD, while non-poor women 
slightly preferred ICRAF. It may therefore be 
concluded that ICRAF made better arrange- 
ments to work with farmers and reached 
them in more sustainable ways via MoARD 
staff, as the government was unable to do 

In ranking methods of dissemination 
employed by external organizations, partic- 
ipants in Mwitubi village identified demon- 
strations and barazas as the most important. 
The poor and non-poor men ranked demon- 
strations at 40 percent, while poor and non- 


poor women ranked barazas at 40 percent. 
On average, however, men appeared to 
prefer demonstrations, whereas women pre- 
ferred the baraza. This is in line with focus 
group discussions on the methods where 
poor men argued that they forgot what they 
learned in barazas and meetings soon after 
learning. This is because lectures do not 
facilitate effective adult learning; hence, the 
participants preferred demonstrations as the 
main means. They also indicated that radio 
was a problematic source of agricultural in- 
formation owing to lack of radios or the bat- 
teries to power them among many farmers. 
Women may have preferred barazas not so 
much of their own free will but because tra- 
ditionally barazas have been compulsory 
for all adults in the villages, and hence saw 
them as an opportunity to fulfill a chief's 
regulation as well as listen to the dissem- 
inators. On the model/contact farmers and 
other farmers frequently involved by the ex- 
ternal organizations in training other farmers, 
the poor men participants were of the view 
that model farmers gained more prestige and 
control over other farmers as they trained 
them. Non-poor men participants argued that 
farmers who trained others gained knowl- 
edge of agricultural technologies earlier and 
interacted better and more easily. They fur- 
ther argued that committees led by women 
were more effective. While poor women ar- 
gued that richer and more educated farmers 
adopted inorganic fertilizers and poorer and 
less educated ones adopted organic prac- 
tices, non-poor women participants were of 
the view that relationships between commit- 
tee members and non-members were amica- 
ble. This was because members reached out 
to non-members with information and the 
latter were eager and willing to learn. Over- 
all, therefore, farmer trainers or committee 
members who were trained to train other 
farmers did not appear to present problems 
to dissemination processes in the village, 
although there were some tensions over the 
contact farmer. 

Looking at the dissemination methods 
at a higher level of generality, participants 

appeared to link the methods to the tech- 
nologies that accompanied them and the tan- 
gible benefits accruing thereto. Poor male 
participants indicated that they liked the 
approaches used to reach farmers in the vil- 
lage, as they were able to train others in their 
groups on SFR shrubs and zero-grazing 
techniques. They were also happy with and 
willing to continue sharing costs for veteri- 
nary services. Non-poor male participants 
liked the technologies in which they were 
trained, as they were beneficial to them. 
Poor women participants liked the dissemi- 
nation approaches because of the training 
activities and the farm inputs that came with 
them. Non-poor women participants, on the 
other hand, liked the dissemination approach 
because of MoARD's training in soil con- 
servation, which had contributed to and 
improved crop production and also because 
the technologies disseminated took a short 
time to implement and realize results. From 
the foregoing, it appears that dissemination 
approaches are better if they deliver tech- 
nologies that relate to the realities of farm- 
ers in ways that are easy and convenient to 

Discussing the type of interaction be- 
tween farmers and the various institutions, 
poor male participants were of the view that 
extension staff were too few and they took 
too long to revisit them after initial training. 
Other groups had a difference experience: 
non-poor men and poor women participants 
indicated that the training offered by the 
various agencies was beneficial to them and 
had been owned by the community. Non- 
poor women participants also thought that 
interaction with MoARD and ICRAF was 
smooth, with the latter using staff from the 
former to reach farmers in the village. Par- 
ticipants also indicated that the local chief 
usually linked farmers to dissemination or- 
ganizations, in particular, MoARD, and took 
initiative to disseminate any knowledge ac- 
quired. However, there was unanimity that 
the extension agencies' policies, practices, 
and training have not responded to farmers' 
feedback and that visits by extension staff 


were scanty, shallow, and too little time was 
often allocated to them. In spite of such 
shortcomings, participants agreed across the 
board that the institutions had developed 
meaningful rapport with local social struc- 
tures such as chiefs, barazas, and farmers' 
groups in general that enabled them to work 
smoothly. Particular farmer participation 
in dissemination processes was wanting. As 
participants in the focus groups indicated, 
farmers participated mainly through re- 
sponding to questions during training ses- 
sions and mobilizing their colleagues for the 
sessions as appropriate. 

As a result of the training provided by 
the disseminating organizations, farmers' 
skills in a variety of technical areas in agri- 
culture had improved. For instance, poor 
women participants said that before ICRAF 
trained them on the use of the shrub tithonia 
for soil fertility improvement, they had con- 
sidered the shrub a weed. Their knowledge 
in using that technology increased from zero 
at the point of intervention to 60 percent at 
the time of this study. Participants said that 
farmer skills in the use of compost, farm- 
yard manure, improved fallows, terraces, in- 
organic fertilizers, and rock phosphates had 
significantly increased following the inter- 
ventions by the dissemination organizations. 

Besides the knowledge farmers gained 
in the technical agricultural areas, partici- 
pants indicated that farmers had developed 
a much closer relationship with each other 
as a result of dissemination work in the 
village. Groups that had thus been set up 
mainly for dissemination resulted in group 
leadership being institutionalized in the vil- 
lage. Overall, study results indicate that in- 
dividuals had gained both technical skills in 
agriculture and social interaction/communi- 
cation skills as they trained their colleagues 
or shared information with them in system- 
atic ways. 

There was consensus among all the four 
groups of participants in the focus group 
discussions that schoolchildren were being 
reached with agricultural dissemination in 
one way or another. Participants said that 

the students learned agriculture in class and 
practiced what they learned both on school 
plots and at their homes. MoARD also 
trained schoolchildren in tree planting 
mostly through the 4K clubs in primary and 
secondary schools. Men participants saw 
the flow of agricultural information from 
school kids to parents and the village in 
general as emanating from the school plots 
and the plots given to the children to prac- 
tice agriculture at home. The men partici- 
pants were of the view that parents and other 
villagers observed those plots and drew 
positive lessons from them. Women partici- 
pants on their part said that students trained 
their parents on the technologies they 
learned in school and also came home with 
agricultural inputs such as tree seedlings 
for planting. All four groups of participants 
were also agreed that technologies dissemi- 
nated to the local school were still there and 
were re-disseminated to the village through 
field days, tree planting days, and songs or- 
ganized in the school. The school can there- 
fore be said to represent a focal point for 
dissemination work in the village. 

Local Organizations and 
Community Relationships 

Prior to intervention, the main groups that 
existed in Mwitubi village included church 
groups, and women's and youth groups as 
well as merry-go-round groups. None of 
the groups mentioned was involved in agri- 
cultural dissemination prior to intervention 
by the external organizations. On the com- 
mencement of intervention in the village, 
some groups were formed to further the ob- 
jectives of dissemination. Through MoARD, 
village committees were formed in 1996. 
The main objective of the committee was 
dissemination of agricultural technologies. 
The participants said that elections to the 
village committee offices were conducted 
when it was formed, members make mone- 
tary subscriptions to it monthly, and meet- 
ings are held ad hoc whenever the leaders 
deem it fit. No initiatives have been taken 
by the committee, as it is weak and had 


almost collapsed when MoARD staff com- 
pleted their intensive backstopping in the 
village and moved to another one the fol- 
lowing year. According to poor female par- 
ticipants, when the committee was active, 
its members (who represented various areas 
of the village) disseminated information to 
villagers through farm trials on committee 
members' farms. Non-poor men for their 
part argued that the committee relied mainly 
on the local chief and their assistant to dis- 
seminate to the villagers. 

All four groups of participants were in 
agreement that village-level committees 
were formed by MoARD for purposes of 
agricultural technologies dissemination. 
The structure of the committees was not 
presented as uniform and while all partic- 
ipants focused on one village, that is, 
Mwitubi, some thought the committee met 
ad hoc; others said it met once, twice, or 
thrice per month. There was also no clear 
understanding by the participants as to how 
the committees disseminated information. 
These contradictions indicate that the vil- 
lage committees have weakened and ceased 
to be major avenues for dissemination in the 
village so that villagers do not remember 
much about them. Village committees in the 
study areas may not have been sustainable 
institutions for agricultural dissemination. 

Participants said that some farmers 
approached the village committee for agri- 
cultural training, but the same was not 
forthcoming. Therefore the objectives of 
the committee were not met. The committee 
was reported as riddled in leadership wran- 
gles to the extent that it scared away the 
members' subscriptions, and MoARD's lim- 
ited visits to the village, given low staffing, 
dealt a final blow to the committee. While 
it lasted therefore, the committee's perfor- 
mance was lackluster at best. In general, the 
main constraints to the work of the village 
committee had to do with leadership prob- 
lems and limited contact with ICRAF and 
MoARD. The village committee did not 
therefore present a dependable avenue for 

dissemination or local organizational change 
and development as would be expected. 
Hence, the committee became more of a 
constraint to social capital formation and 
human capital development/empowerment 
in the village. 

The work of the committee appears to 
have gone smoothly but, without design, 
moved into the hands of the administration, 
that is, the chiefs and their assistant. While 
participants reported that the administration 
acted in consultation with the committee 
in mobilizing farmers for dissemination 
sessions and doing actual dissemination in 
barazas, by the time ICRAF arrived, they 
worked more closely and directly with the 
chiefs and their assistants in mobilizing 
farmers and disseminating to them. It is for 
that reason that poor women participants 
reported that chiefs and councilors were 
active in committee activities and in dissem- 
ination to non-committee members, adding 
"farmers attended dissemination meetings 
organized by chiefs and their assistants." 
Underscoring the point further, non-poor 
women noted, "The chief is the unifying 
factor in the village, linking the community 
with (dissemination) service providers." 
There is need to point out that beyond the 
village where the committee had not per- 
formed well, the catchment committee, 
which was wider, was reported to have 
achieved highly in soil conservation efforts 
in the same community but at a higher level. 

In general, and beyond village commit- 
tee activities, participants were of the view 
that dissemination efforts facilitated cooper- 
ation among farmers in Mwitubi village. 
There were no differences in the views of 
the various groups of participants on how 
community cooperation/conflict relates to 
the dissemination approaches. All the groups, 
of men and of women, of the poor and of the 
rich, appear to agree on the fact that the ex- 
tension interventions led to cooperation and 
cohesion in some ways and to competition 
and conflict in other ways. Overall, the par- 
ticipants seem to have viewed the emerging 


conflicts/competition more as learning points 
than as negative sides of extension. This 
is perhaps why one group of participants 
looked beyond the life of a dissemination 
project to argue that farmers interact closely 
even after the external disseminating organ- 
ization had left. These, therefore, are the ef- 
fects of the practices of external institutions 
on local social relationships. 

Participants reported that their main con- 
tribution to dissemination efforts constituted 
their mobilizing other farmers to attend dis- 
semination sessions and to the in-kind con- 
tributions, for example, in terms of plots 
for experimentation. In terms of sharing the 
costs of extension work, non-poor men said 
they were unwilling, while poor women 
said they were willing, so long as they were 
empowered financially through some credit 
scheme. However, the main bases of sus- 
tainability of the dissemination approach 
and practices lay in the involvement of the 
chiefs, their assistants, and village elders in 
mobilizing farmers and often disseminating 
directly to them. All the groups of partici- 
pants recognized the chiefs and their capac- 
ity to mobilize farmers as critical to the sus- 
tainability of the dissemination approaches. 

Poor women participants added another 
dimension to the sustainability, arguing that 
the fact that individual farmers appreciated 
the critical place of dissemination in their 
work was important. This is because such a 
scenario creates a demand-led dissemination 
process that is less likely to collapse than a 
situation where farmers had not fully recog- 
nized the need for extension. From the fore- 
going, however, the dissemination approach 
and process in Mwitubi village can be said to 
be sustainable only if the chiefs, their assis- 
tants, and the village elders had been trained 
as trainers of trainers. In the event that none 
of the disseminating organizations trained 
them, as is the case, they have no knowledge 
to disseminate and therefore there is need 
for them to have been linked with trained 
farmers in order to close the knowledge gap 
and sustain the process and approach. 


The dissemination organizations generally 
satisfied farmers' expectations of learning 
new or improved agricultural technologies. 
The organizations provided training on 
relevant and innovative technologies that 
interested the farmers. The organizations 
did not study the communities they worked 
with at the commencement of their projects, 
however, to understand them and their prob- 
lems fully. MoARD, for instance, moved 
into the village and formed the catchment/ 
village committees, ignoring local institu- 
tions and groups by which people organized 
their agricultural activities. The farmers 
argued that through that approach, MoARD 
contributed significantly to soil conservation 
in the village. The success of the approach 
is best explained, however, by the impor- 
tance of the problem it tackled rather than 
the efficacy of the approach itself. Indeed, 
when ICRAF arrived in the village, farmers 
led its workers more to the baraza and the 
chiefs than to the catchment committee as 
the focal point for dissemination in the 

ICRAF and MoARD did not as a matter 
of policy proceed to involve farmers in their 
dissemination processes. Study results show 
no evidence of farmer participation in the 
development of dissemination approaches, 
let alone their fine-tuning over time. Farmer 
involvement was limited even in the devel- 
opment of training materials. Farmers said 
that their feedback did not influence agency 
policies and practices. While the organiza- 
tions appear to have developed high levels 
of rapport with farmers and their groups to 
the extent that their dissemination meetings 
were well attended and supported by local 
leaders such as chiefs and village heads, crit- 
ical participation that could empower farm- 
ers to be self-reliant afterwards was lacking. 
Farmers or committees do not appear to 
have followed the dissemination processes 
and internalized them to a level that they 
may broker such services on their own in the 
future. Discussions by the various groups of 


participants brought out very clearly the 
critical role of schools in the dissemination 
process. Schoolchildren were said to come 
home not just with skills but also with in- 
puts such as seedlings, which they planted 
at home. In general, the children attempted 
to train their parents and other farmers on 
what they learned. Fine-tuned properly, the 
school approach to agricultural dissemina- 
tion could become a very important avenue, 
because the school kids were very moti- 
vated to learn and implement. The children 
need, however, to be facilitated to get plots 
and inputs on which to practice and from 
there influence their parents and communi- 
ties. The trees already planted in the village 
school with the help of MoARD are im- 
portant and may be replicated for demon- 
stration, school income, and observation by 
farmers adjacent to the schools. 

With regard to local leadership and insti- 
tutions, chiefs, their assistants, and village- 
level leaders were critical in the dissemina- 
tion processes. They mobilized farmers at 
no cost and to the convenience of the dis- 
seminators. While the organizations fell into 
the trap of village life as organized around 
the chief's baraza, however, farmers forgot 
most of the dissemination messages given 
in the barazas almost instantly in their own 
admission. Farmers preferred demonstra- 
tions and farm trials that provided them 
with opportunities for practical learning. 

The Umbrella Group 
Approach in Bukhalalire 

Bukhalalire village is in Bukhalalire sub- 
location, Malachi East Location, Butula Di- 
vision of Busia District, Western Province. 
The Luhya occupy the village. These are 
mainly small-scale resource-limited farmers 
with low literacy levels. Like in other parts 
of the country, agricultural dissemination in 
the village was pioneered by MoARD dur- 
ing the colonial period. However, during 
the era covered by this study, dissemination 
was through the catchment area approach, 
which was started in 1993 by MoARD. 

When Kenya Woodfuel Agroforestry Pro- 
gramme (KWAP) entered the village in 
1994, the umbrella group approach was 
adopted. While the former approach fo- 
cused on soil and water conservation as they 
affected people living within a common 
drainage system, the latter focused more on 
agroforestry and other related agricultural 
technologies. The umbrella group brought 
together about 1 1 groups in the villages that 
were interested in agricultural dissemination. 
They elected two representatives each to the 
executive committee. These were trained 
and they in turn trained members of the 
sponsoring groups. The chiefs were critical 
in the dissemination process as they con- 
vened barazas and mobilized farmers, most 
of whom were not members of the groups to 
attend the dissemination meetings. 

Farmers' Comparative Evaluation 
of the External Organizations 

Focus-group discussion participants reported 
that several organizations had worked in the 
village, covering a number of issues. These 
organizations included MoARD, KWAP, 
KARI, Tropical Soils Biology and Fertility 
Programme (TSBF), and ICRAF, but the 
presence of the first two was much stronger 
than the other organizations. Agriculture was 
reported to be the central theme of all the 
organizations' work (there was an additional 
organization, KENFINCO, that focused on 
the provision of water to the villagers). 

When participants were asked to rank 
the various external organizations that had 
disseminated in the village, ICRAF was 
rated at 50 percent, 20 percent, 22 percent, 
and 13 percent by poor men, poor women, 
non-poor men, and non-poor women, re- 
spectively. On similar lines, poor men, poor 
women, non-poor men, and non-poor women 
rated KARI at 25 percent, 40 percent, 33 
percent, and 20 percent and MoARD at 25 
percent, 40 percent, 44 percent, and 66 per- 
cent, respectively. Both men and women par- 
ticipants ranked MoARD higher, on average, 
than all the other organizations. This perhaps 
results from the long history of associating 


agricultural dissemination in Kenya with 
MoARD as it was for a long time the sole 
service provider. Comparing the poor and 
non-poor participants closely indicates that 
the poor ranked ICRAF and CARE higher, 
while the non-poor ranked MoARD higher. 
This may be a result of ICRAF's and 
CARE's focus on bio-intensive, low-capital 
requiring and labor-intensive agriculture 
that resonates with the circumstances of the 
poor, compared to MoARD's focus on high- 
capital requiring and external-input-based 
agriculture that the non-poor readily iden- 
tify with. 

The foregoing point also relates to the 
differences between men and women in 
perceived benefits from the organizations. 
While men participants argued that adop- 
tion was slow and dis-adoption was setting 
into the village, because KARI did not ful- 
fill its promises to the farmers and TSBF 
had focused on only two farmers, women 
maintained that disseminated technologies 
were beneficial and were taken very seri- 
ously by farmers. They associated most 
good developments in the community with 
KWAP, saying, "Schools and homes alike 
have income from the trees planted earlier 
and the tree nurseries established on a con- 
tinuous basis." 

Overall, all the groups of participants 
tended to agree on the fact that for tech- 
nologies to be useful to the community, they 
had to be relevant to local needs. They iden- 
tified KWAP-disseminated technologies as 
most relevant and therefore most helpful. 
Participants also concurred that the um- 
brella group approach was most appropri- 
ate because it worked through groups and 
reached more farmers for training. Partici- 
pants' views did not contradict each other 
on the basis of gender or class. 

When participants ranked the various 
methods employed by the external institu- 
tions in dissemination, demonstrations and 
meetings/barazas were identified as the 
most important methods. On average, both 
men and women preferred demonstrations 
to the same level of 20 percent and, hence, 

preference of the method did not vary with 
gender. On balance, however, women ranked 
meetings/ barazas higher than their men 
counterparts did. In the focus group discus- 
sions, poor men and poor women partici- 
pants were agreed that training did not reach 
all the farmers that it should have reached. 
All the groups were also agreed that the 
departure of the disseminating organization 
before the agreed time could be one fac- 
tor to blame. Although poor men thought 
that the use of barazas as venues for train- 
ing could increase coverage, rich men and 
women were of the view that training pro- 
vided was sufficient but that it could be 
improved through increased training as well 
as monitoring and evaluation by the organi- 
zations. Overall, the barazas/mee tings ap- 
pear to have been preferred by women and 
the poor, possibly resulting from the weak 
social position of women and the poor in the 
Luhya society and, hence, their propensity 
to obey regulations such as attending barazas 
on chiefs' orders. 

Besides the barazas/meetings, umbrella 
group committee members had the respon- 
sibility of reaching out to farmers within 
and outside the catchment. Discussing 
possible increase of skills and confidence of 
individual farmers, poor men participants 
said that umbrella group members are oc- 
casionally invited by outside groups and 
schools to give seminars and briefings on 
agriculture. This implies that some farmers 
had improved their communication skills as 
well as acquired knowledge in forms that 
they could individually and independently 
disseminate. Non-poor men participants also 
argued that umbrella group members were 
usually consulted by other farmers on vari- 
ous technical areas and they trained other 
farmers even beyond the catchment; hence, 
their confidence had improved. Poor and 
non-poor women participants expressed sim- 
ilar views on the capacity of umbrella group 
members to disseminate within and without 
the catchment. This implies that farmer 
trainers were effective and had become a 
significant component of the dissemination 


approach in Bukhalalire village. In ranking 
the internal organizations involved in dis- 
semination in Bukhalalire village, partici- 
pants identified barazas and other farmers 
as major sources of information. Non-poor 
participants ranked both barazas and other 
farmers higher. On average, while women 
preferred barazas, men identified more 
with other farmers. This scenario is borne 
out by the fact that men among the Luhya 
have more time on their hands to visit each 
other than women, who do most farmwork 
and almost all domestic chores. Women 
then attend barazas both as a requirement 
of the chief and also to get agricultural in- 
formation. Therefore, the main currents of 
dissemination in the village coalesce 
around farmer-to-farmer communication 
and barazas. 

It is for the foregoing reasons that the 
chief who convenes barazas was a central 
subject of discussion when participants 
looked at communication linkages that sup- 
ported dissemination. Therefore, when the 
participants looked at committee interactions 
with and demands on local authorities or 
chiefs, poor men said the umbrella group 
had links with the local administration, 
mainly the chiefs, who provide them with 
security and help with dissemination 
through barazas. They said that farmers 
send their chiefs to negotiate demands with 
MoARD. Dissemination organizations, 
including MoARD, KWAP, KARI, and 
ICRAF, all came in through the chief's 
baraza. Non-poor men participants argued 
that there existed a strong link between the 
umbrella group, the chiefs, and their assis- 
tant at the time of entry of a dissemination 
organization into the village, for mobiliza- 
tion, and actual dissemination, on occasion. 
Poor women and non-poor women partici- 
pants concurred with the foregoing that 
chiefs and their barazas had been critical to 
dissemination work in the village. 

On farmers' knowledge of their rights, 
from the poor men participants' perspective, 
it would appear that they knew their rights 
to extension. Non-poor men and poor 

women were of the view that farmers knew 
what they needed from each of the dis- 
seminating organizations and made efforts 
to make demands on them. To this extent, 
they therefore pursued their rights to exten- 
sion services. Non-poor women saw no 
evidence of farmers' knowledge of their 
rights. The participants were agreed that 
farmers had increased their skills and 
knowledge as a result of the dissemination 
efforts undertaken. Poor men participants 
said farmers had increased skills in tree nurs- 
ery establishment and tree planting, which 
they learned from KWAP, as well as SFR 
technologies, which they learned mainly 
from ICRAF. Non-poor men for their part 
mentioned increased skills in soil conser- 
vation and fertility improvement. Poor and 
non-poor women mentioned fodder crops, 
and soil and water conservation. Besides the 
technical skills that translated into better- 
conserved, fertile soils and higher farm 
yields, farmers had also acquired interper- 
sonal communication skills, which enabled 
them to exchange knowledge in agriculture 
more effectively. They had come to a level 
in which they recognized each other as 
sources of useful agricultural information. 

The school program was strong in 
Bukhalalire village. The participants identi- 
fied a variety of programs that were target- 
ing schools with dissemination activities. 
Poor men identified school demonstration 
farms. Non-poor men and non-poor women 
for their part identified the CARE program 
working in conjunction with Action Aid at 
Bukhalalire primary and secondary schools. 
They added that KWAP also worked with the 
schools, although not directly but through 
the umbrella group. Poor women said that 
the KWAP program assisted Bukhalalire 
primary school to establish a woodlot. 

Asked whether the technologies were 
reaching the schools, poor men participants 
argued that schoolchildren both at primary 
and secondary levels learn about various 
technologies and have practiced them at 
home in the past. Non-poor men said that 
the technologies reached the schools and 


children of about three generations ago 
could disseminate what they had learned 
from school by having their own farms at 
home, but such is not going on now. They 
added that KWAP had a woodlot at the vil- 
lage school. Poor women concurred with 
the foregoing, naming the trees in the wood- 
lot as including grevillea, casuarina, and 
jacaranda. Non-poor women said the wood- 
lot was still in the village school and that the 
trees had not benefited the school as they 
have not matured for harvest. They added 
that at present, the children learn a great 
deal of agricultural theory in school with 
little practical application. 

When participants discussed how and if 
students trained their parents and other vil- 
lagers on what they had learned in school, 
non-poor participants were of the view that 
children did not undertake any training. 
Poor participants for their part argued that 
students tried to train parents as well as 
demonstrate at home what they had learned 
in school. The implication of the foregoing 
is that the poor have limited sources of tech- 
nical information and hence are keen to re- 
ceive information from any source, includ- 
ing their children, while the non-poor have 
options and may not emphasize their chil- 
dren as a source. 

Some participants said that the technolo- 
gies disseminated to schools were still being 
utilized, adding that the woodlot established 
at Bukhalalire primary school was benefiting 
the school with income through the sale of 
trees. When the participants turned to prob- 
lems associated with the schools programs, 
poor men indicated that for whatever rea- 
son, pupils no longer practice what they learn 
in school at their homes, while non-poor 
men identified no problems with the pro- 
grams. Poor women participants for their 
part said that the major problem was that 
pupils try to reach out to their parents with 
the technologies that they learn in school 
but they hardly convince them to adopt. The 
non-poor women said that there was no prac- 
tical learning for kids in schools and, hence, 
the children have little to disseminate. They 

added that cattle from the school's neigh- 
borhood were destroying woodlots in the 

Local Organizations and 
Community Relationships 

On organizations and informal networks 
that existed prior to intervention, poor men 
participants said that there were women's, 
youth, clan, and merry-go-round groups and 
of those, youth groups were said to provide 
farm labor at a fee to raise revenue. Non- 
poor men added church groups to the poor 
men's list, adding that Kuku women's group 
was engaged in agricultural activities. Other 
focus groups identified others, like a clay 
work group and church groups, but they 
were involved in what was considered non- 
agricultural activities. 

When discussing the new groups or 
committees formed through the interven- 
tions, poor men mentioned Kuku women's 
group and Bukhalalire handcraft women's 
group. They said that the two groups were 
part of the umbrella group and their foremost 
objective was agricultural dissemination to- 
gether with poultry keeping and handcraft 
business, respectively. In terms of organiza- 
tion, the groups are affiliated to the umbrella 
group committee, which coordinates activi- 
ties in the entire catchment. The groups elect 
their own representatives to the umbrella 
group committee. The women's groups are 
mostly informal. The umbrella group under- 
takes local training and dissemination out- 
side the catchment in other groups and in 
schools. Non-poor men indicated that the 
new groups include women's groups, which 
became part of the umbrella group. These 
women's groups included Amuka, Namudu 
Wekhonye, and Banguria, and all are in- 
volved in dissemination as part of the um- 
brella group committee. Organizationally, 
they said that the groups are from various 
villages and are represented in the umbrella 
group committee by two representatives 
each. Elections to the umbrella group com- 
mittee were supposed to be held annually, 
but an arrangement was struck to hold them 


after every three years. The committee was 
involved in dissemination within and out- 
side the catchment and in schools through 
field days, informal oral discussions, and ob- 
servation. Poor women concurred with the 
non-poor men on the new groups set up fol- 
lowing intervention, how they operate, and 
were organized. 

Non-poor women on their part said that 
following intervention, the umbrella group 
was set up and 1 1 groups sent two represen- 
tatives each to form the executive commit- 
tee of the group. After the umbrella group 
was formed, the main objective of all 
groups represented in the executive com- 
mittee became to disseminate agricultural 
technologies. The umbrella group usually 
met once each month and, in case of urgent 
matters to deal with, they met more often. 
Participants said that the attendance in the 
meetings was good. The participants also 
pointed out that elections to the umbrella 
group executive committee were supposed 
to be held after three years and those elected 
at the start of the group were reelected for 
a second term. They further said that the 
umbrella group is useful in dissemination 
through groups' demonstrations and field 
days; neighboring villages learn from them 
informally and through attendance in dis- 
semination sessions. 

Participants were agreed that the um- 
brella group approach was effective in reach- 
ing farmers with dissemination messages. 
It is for that reason that participants were of 
the view that farmers had been involved in 
developing training materials and that the 
groups had continued with dissemination 
work. When participants discussed the in- 
volvement of committees in further dissem- 
ination outside the village, poor men said 
that umbrella groups have been involved in 
developing training materials and are occa- 
sionally invited outside the catchment to 
give seminars and briefings on agriculture 
to schools. They also reach out to other vil- 
lages to undertake dissemination through 
informal contacts or by invitation. Non-poor 

men said that the umbrella group undertakes 
dissemination to other villages through in- 
formal oral discussions and field days. Some 
of their members attend field days in neigh- 
boring catchments as well as villages and 
schools and their neighbors in turn attend 
theirs. Poor women said that non-members 
of the umbrella groups attend and learn in 
the field days, which they organize without 
outside staff. Non-poor women concurred 
with the foregoing. 

The participants further indicated that 
the groups were involved in new dissemina- 
tion activities after the external organizations 
had left. These activities were undertaken 
because the methods of dissemination in use 
had been internalized by the groups and 
were being replicated independently by the 
community and its structures. Therefore, on 
new activities undertaken by the dissemina- 
tion groups, poor men participants said that 
the umbrella group has given briefings to 
schools about their activities and reached 
out to disseminate to other villages outside 
the catchment. Non-poor men participants 
mentioned exchange visits to neighboring 
villages and catchments as well as field 
days on farms where technologies had been 
implemented. The instructors in the field 
days were drawn from the umbrella groups. 
Poor women and non-poor women said they 
had seen no new activities undertaken by 
the groups. Although effective, however, the 
methods in use and the activities engen- 
dered, especially after the external orga- 
nizations had left, appear to have excluded 
women or proceeded in ways that did not 
capture women's imaginations, so that they 
did not notice them. 

In considering the effect of the umbrella 
group approach to dissemination on com- 
munity power structure, participants across 
the spectrum agreed that the approach had 
brought community members closer to- 
gether in "discussing and exchanging in- 
formation about the various technologies." 
The participants also indicated that benefits 
from the group did not depend on the status 


of a member and that people benefited from 
the group equally, although implementation 
of technologies depended ultimately on how 
hard one works. Therefore, the umbrella 
group approach to dissemination had con- 
tributed significantly to the evolution of a 
cohesive community that strove to be self- 
reliant through intra-community exchange 
of information. 


When participants discussed the status of 
the groups, poor men said that the umbrella 
group was still in existence for the purpose 
of continuing to benefit from the technolo- 
gies disseminated. They argued, "the um- 
brella group continues to disseminate to 
other villages beyond the catchment and 
giving briefings on their activities to 
schools." Non-poor men participants said 
that the umbrella group is still in existence 
and members meet once per month, but if 
there is an urgent matter, meetings can be 
called at any time. They noted, "in the meet- 
ings they discuss ongoing activities and how 
to enhance adoption of the technology that 
they disseminate." They further said that 
currently, umbrella group members are in- 
volved in dissemination through field days, 
oral discussion, and observation on farms 
where implementation has taken place. 

Poor women participants said that the 
umbrella group was still in existence and 
that members make contributions and have 
participated in mobilizing farmers for field 
days. The participants said that the umbrella 
group was continuing with dissemination, 
arguing, "up to now they organize field 
days with very little support from external 
dissemination agencies." Non-poor women 
participants concurred with the foregoing, 
reporting that the umbrella group was or- 
ganizing field days and getting in touch 
with disseminating organizations that came 
after KWAP to increase their pool of knowl- 
edge. They also said that the catchment com- 
mittee was active in water and soil conserva- 
tion, although not as active as during the time 

before the formation of the umbrella group. 
They concluded that the umbrella group was 
continuing with dissemination both within 
and outside the catchment. 

Poor men participants also indicated that 
dissemination activities were continuing in 
the catchment and umbrella group members 
have been involved in doing it in villages 
and schools outside the catchment. Farmers 
from outside also attend trials and other ac- 
tivities in the catchment. Non-poor men said 
that the umbrella group has had contact with 
other outside agencies after KWAP left and 
this has kept the group working on benefi- 
cial technologies even after the former left. 
This is because making contact with other 
dissemination agencies helps replenish their 
know-how and encourages them to keep 
working on dissemination. Poor women 
participants concurred with the foregoing. 
Non-poor women for their part argued that 
non-group members are reached through 
the field days, which they are free to attend, 
and informally as they observe and discuss 
with group members. They also indicated 
that umbrella group members have also 
gained knowledge and planting materials 
(mainly cassava) from outside the catch- 
ment, especially from friends and relatives. 


All groups of participants recognized that 
selection of technologies for dissemination 
should be linked very closely with the needs 
of the villagers. For this reason, the tech- 
nologies disseminated by KWAP resonated 
well with them. Participants said that 
farmers' needs for cash, building materials, 
and fuelwood were readily met as a result of 
the technologies disseminated by KWAP. 
Community members were therefore eager 
to implement those technologies. Farmers 
also learned from each other easily because 
the technologies met their felt needs. 

Participants also thought that the barazas 
and, hence, the chiefs and their assistant 
were critical to dissemination processes in 
the village. They identified with the baraza 


as their meeting point for dissemination and 
sent the chief to MoARD with demands for 
extension services. They therefore recog- 
nized the chief as a knowledge focal point 
and implied that further training of the chief 
in the area of dissemination of agricultural 
technologies would be beneficial. Farmer- 
to-farmer communication of agricultural 
technologies was also strong, especially 
among men in the village. 

All participants also recognized the um- 
brella approach as effective in the village. 
They argued that it involved farmers from 
a variety of groups and was therefore able 
to reach more farmers than was possible 
with other approaches. They viewed the ap- 
proach of using existing groups for exten- 

sion as more inclusive and for that reason 
criticized other approaches, such as those by 
TSBF that focused on a few individuals. 

The groups and the dissemination ac- 
tivities they promoted were also judged as 
sustainable. Farmers' groups have increased 
and have come to focus on dissemination as 
their main purpose. These groups have also 
continued with dissemination work long 
after the external institutions that worked 
with them have left the community. In addi- 
tion, the groups have reached out with agri- 
cultural information to farmers within and 
outside the village where work in collabo- 
ration with external organization had origi- 
nally started. 


Arce, A. 1993. Negotiating agricultural development. Entanglements of bureaucrats and rural producers 
in western Mexico. Wageningen Studies in Sociology, Wageningen University, The Netherlands. 

Argwings-Kodhek, G., T. Jayne, G. Nyambane, T. Awuor, and T. Yamano. 1999. How can micro-level 
household information make a difference for agricultural policy making? Tegemeo Institute of 
Agricultural Policy and Development, Kenya Agricultural Research Institute, and Michigan 
State University. Photocopy. 

Behrman, J. R., H. P. Kohler, and S. Cotts Watkins. 2002. Social networks and changes in contraceptive 
use over time: Evidence from a longitudinal study in rural Kenya. Demography 39 (4): 713-738. 

Carney, D., ed. 1998. Sustainable rural livelihoods: What contributions can we make? Department for 
International Development, London. 

Desaeger, J., and M. R. Rao. 1999. The root-knot nematode problem in Sesbania fallows and scope for 
managing it in western Kenya. Agroforestry Systems 47: 273-288. 

De Wolf, J., R. Rommelse, and A. Pisanelli. 2000. Improved fallow technology in western Kenya: Po- 
tential and reception by farmers. Nairobi: International Centre for Research in Agroforestry. 

Ellis, F. 2000. Rural livelihoods and diversity in developing countries. Oxford: Oxford University Press. 

Hartemink, A. E., R. J. Buresh, B. Jama, and B. H. Janssen. 1996. Soil nitrate and water dynamics in 
Sesbania fallows, weed fallows and maize. Soil Science Society of America Journal 60: 568-574. 

Hebinck, P., and N. Mango. 2001. Kinship relations and land inheritance among the Luo. In Kinship 
structures and enterprising actors. Anthropological essays on development, eds. J. Andersson 
and M. Breusers. Sociology of Rural Development, Wageningen University, The Netherlands, 
pp. 37-57. 

IFPRI (International Food Policy Research Institute). 2000. Impact of agricultural research on poverty re- 
duction: An integrated economic and social analysis. Phase II. Proposal prepared on behalf of the 
IAEG/SPIA by the International Food Policy Research Institute. June 8, 2000, Washington, D.C. 

Jama, B., R. A. Swinkels, and R. J. Buresh. 1997. Agronomic and economic evaluation of organic and 
inorganic sources of phosphorus in western Kenya. Agronomy Journal 89: 597-604. 

Jama, B., R. J. Buresh, and F. M. Place. 1998a. Sesbania tree fallows on phosphorus-deficient sites: 
Maize yield and financial benefit. Agronomy Journal 90 (6): 717-726. 

Jama, B., A. Niang, J. De Wolf, B. Amadalo, and M. R. Rao. 1998b. Sources of nutrients in nutrient de- 
pleted soils of western Kenya. Proceedings of CIMMYT's Sixth Regional Maize Conference for 
Eastern and Southern Africa, Addis Abeba, Ethiopia, September 21-25, 1998. 

Jayne, T S., J. Strauss, T. Yamano, and D. Molla. 2001. Giving to the poor? Targeting of food aid in 
rural Ethiopia. World Development 29 (5): 887-910. 

Kenya — Ministry of Planning and Finance. 2000. Second report on poverty in Kenya: Incidence and 
depth of poverty, Vol. 1. Nairobi. 

Kwesiga, F, and R. Coe. 1994. The effect of short rotation Sesbania sesban planted fallows on maize 
yield. Forest and Ecology 44: 403^28. 



Leonard, D. K. 1991. African successes: Four public managers of Kenyan rural development. Berke- 
ley: University of California Press. 

Long, N. 2002. Development sociology. Actor perspectives. London: Routledge. 

Mango, N. 2002. Husbanding the land. Agrarian development and socio-technical change in Luoland, 
Kenya. Ph.D. thesis. Wageningen University, the Netherlands. 

Meinzen-Dick, R., M. Adato, L. Haddad, and P. Hazell. 2004. Science and poverty: An interdisciplinary 
assessment of the impact of agricultural research. Food Policy Report. Washington, D.C.: Inter- 
national Food Policy Research Institute. 

Niang, A., B. Amadalo, M. Adongo, and J. De Wolf. 1996. Weed dynamics under improved fallow sys- 
tems. Poster at the International Symposium on the Science and Practice of Short-Term Improved 
Fallows in Humid and Sub-Humid Tropics, Lilongwe, Malawi, March 11-13, 1996. 

Niang, A., J. De Wolf, S. Gathumbi, and B. Amadalo. 1999. The potential of short duration improved 
fallow with selected trees and shrubs for crop productivity enhancement in the densely popu- 
lated highlands of western Kenya. In La Jachere en Afrique Tropicale: Roles, Amenagements, 
Alternatives. Dakar, April 13-16, 1999. 

Omiti, J., and M. Omosa. 2002. The impact of HIV/AIDS on rural livelihoods in Kenya: A desk review. 
Nairobi and London: AIDS Control Unit, Ministry of Agriculture and Rural Development, U.K. 

Omosa, M. 1998. Re-conceptualizing food security. Interlocking strategies, unfolding choices and rural 
livelihoods in Kisii Dictrict, Kenya. Ph.D. thesis. Sociology of Rural Development, Wageningen 
University, the Netherlands. 

Onim, J. F M., K. Otieno, and B. Dzowela. 1990. The role of Sesbania as multi-purpose tree in small- 
scale farms in western Kenya. In Perennial Sesbania species in agroforestry systems, eds. 
B. Macklin and D. O. Evans. Proceedings of a workshop held in Nairobi, March 27-31, 1989. 
The Nitrogen Fixing Tree Association, Hawaii, pp. 167-179. 

Oswald, A., H. Frost, J. K. Ransom, J. Kroschel, K. Shepherd, and J. Sauerborn. 1996. Studies on poten- 
tial for improved fallow using trees and shrubs to reduce Striga infestations in Kenya. In Advances 
in parasitic plant research, eds. M. T. Moreno, J. I. Cubero, D. Bemer, D. Joel, L. J. Musselman, 
and C. Parker. Proceedings of the Sixth Parasitic Weed Symposium, Cordoba, Spain, pp. 795-800. 

Rhoades, R. 1998. Participatory watershed research and management: Where the shadow falls. Gate 
Keepers Series No. 18. London: International Institute for Environment and Development. 

Rodgers, J. L., H.-P Kohler, J. R. Behrman, and S. C. Watkins. 2001. The density of social networks 
and fertility decisions: Evidence from South Nyanza District, Kenya. Demography 38 (1): 43-58. 

Scoones, I. 1998. Sustainable rural livelihoods: A framework for analysis. Working Paper No. 72. 
Brighton, U.K.: Institute of Development Studies, University of Sussex. 

Shepherd, K. D., E. Ohlsson, J. R. Okalebo, and J. K. Ndufa. 1996. Potential impact of agroforestry on soil 
nutrient balances at the farm scale in the East African Highlands. Fertilizer Research AA: 87-99. 

Stoorvogel, J., and E. Smaling. 1990. Assessment of soil nutrient depletion in sub-Saharan Africa: 
1983-2000. Report No. 28, \-A. Wageningen, Netherlands: Winand Staring Center. 

Swinkels, R., S. Franzel, K. D. Shepherd, E. Ohlsson, and J. K. Ndufa. 1997. The economics of short 
rotation improved fallows: Evidence from areas of high population density in western Kenya. 
Agricultural Systems 55: 99-121. 

Villarreal, M. 1994. Wielding and yielding: Power, subordination and gender identity in the context of 
a Mexican development project. Ph.D. thesis. Sociology of Rural Development, Wageningen 
University, the Netherlands. 

Wallman, S. 1984. Eight London households. London: Routledge. 

Wangila, J., R. Rommelse, and J. De Wolf. 1999. An insight into the socioeconomics of agroforestry 
client-farmers in western Kenya. Pilot Project Report No. 12. Nairobi: International Centre 
for Research in Agroforestry.